re PR target/60610 (ICE in convert_regs_1, at reg-stack.c:3064)
[gcc.git] / gcc / ChangeLog
1 2014-03-21 Jakub Jelinek <jakub@redhat.com>
2
3 PR target/60610
4 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
5 redefine to 1 or 0.
6 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
7 TARGET_ISA_64BIT_P(x).
8
9 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10
11 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
12 pattern for vector nor instead of subtract from splat(-1).
13 (altivec_expand_vec_perm_const_le): Likewise.
14
15 2014-03-21 Richard Henderson <rth@twiddle.net>
16
17 PR target/60598
18 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
19 related insns after epilogue_completed.
20
21 2014-03-21 Martin Jambor <mjambor@suse.cz>
22
23 PR ipa/59176
24 * cgraph.h (symtab_node): New flag body_removed.
25 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
26 when removing bodies.
27 * symtab.c (dump_symtab_base): Dump body_removed flag.
28 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
29 had their bodies removed.
30
31 2014-03-21 Martin Jambor <mjambor@suse.cz>
32
33 PR ipa/60419
34 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
35 in the border.
36
37 2014-03-21 Richard Biener <rguenther@suse.de>
38
39 PR tree-optimization/60577
40 * tree-core.h (struct tree_base): Document nothrow_flag use
41 in VAR_DECL_NONALIASED.
42 * tree.h (VAR_DECL_NONALIASED): New.
43 (may_be_aliased): Adjust.
44 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
45
46 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
47
48 * expr.c (expand_expr_real_1): Remove outdated comment.
49
50 2014-03-20 Jakub Jelinek <jakub@redhat.com>
51
52 PR middle-end/60597
53 * ira.c (adjust_cleared_regs): Call copy_rtx on
54 *reg_equiv[REGNO (loc)].src_p before passing it to
55 simplify_replace_fn_rtx.
56
57 PR target/60568
58 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
59 into CONST, put pic register as first operand of PLUS. Use
60 gen_const_mem for both 32-bit and 64-bit PIC got loads.
61
62 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
63
64 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
65
66 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
67
68 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
69 around for store forwarding issue in the FPU on the UT699.
70 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
71 loads and operations if -mfix-ut699 is specified.
72 (divtf3_hq): Tweak attribute.
73 (sqrttf2_hq): Likewise.
74
75 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
76
77 * calls.c (store_one_arg): Remove incorrect const qualification on the
78 type of the temporary.
79 * cfgexpand.c (expand_return): Likewise.
80 * expr.c (expand_constructor): Likewise.
81 (expand_expr_real_1): Likewise.
82
83 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
84
85 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
86 of parts.
87
88 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
89
90 PR target/60039
91 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
92
93 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
94
95 * config/arm/aarch-common-protos.h
96 (alu_cost_table): Fix spelling of "extend".
97 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
98
99 2014-03-19 Richard Biener <rguenther@suse.de>
100
101 PR middle-end/60553
102 * tree-core.h (tree_type_common): Re-order pointer members
103 to reduce recursion depth during GC walks.
104
105 2014-03-19 Marek Polacek <polacek@redhat.com>
106
107 PR sanitizer/60569
108 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
109 before accessing it.
110
111 2014-03-19 Richard Biener <rguenther@suse.de>
112
113 PR lto/59543
114 * lto-streamer-in.c (input_function): In WPA stage do not drop
115 debug stmts.
116
117 2014-03-19 Jakub Jelinek <jakub@redhat.com>
118
119 PR tree-optimization/60559
120 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
121 with build_zero_cst assignment.
122
123 2014-03-18 Kai Tietz <ktietz@redhat.com>
124
125 PR rtl-optimization/56356
126 * sdbout.c (sdbout_parms): Verify that parms'
127 incoming argument is valid.
128 (sdbout_reg_parms): Likewise.
129
130 2014-03-18 Richard Henderson <rth@redhat.com>
131
132 PR target/60562
133 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
134 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
135 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
136
137 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
138
139 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
140 * doc/plugins.texi (Plugin callbacks): Mention
141 PLUGIN_INCLUDE_FILE.
142 Italicize plugin event names in description. Explain that
143 PLUGIN_PRAGMAS has no sense for lto1. Explain
144 PLUGIN_INCLUDE_FILE.
145 Remind that no GCC functions should be called after
146 PLUGIN_FINISH.
147 Explain what pragmas with expansion are.
148
149 2014-03-18 Martin Liska <mliska@suse.cz>
150
151 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
152 gimple call statement is update.
153 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
154 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
155
156 2014-03-18 Jakub Jelinek <jakub@redhat.com>
157
158 PR sanitizer/60557
159 * ubsan.c (ubsan_instrument_unreachable): Call
160 initialize_sanitizer_builtins.
161 (ubsan_pass): Likewise.
162
163 PR sanitizer/60535
164 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
165 varpool_finalize_decl instead of rest_of_decl_compilation.
166
167 2014-03-18 Richard Biener <rguenther@suse.de>
168
169 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
170 by using bitmap_and_compl instead of bitmap_and_compl_into.
171 (df_rd_transfer_function): Likewise.
172
173 2014-03-18 Richard Biener <rguenther@suse.de>
174
175 * doc/lto.texi (fresolution): Fix typo.
176
177 2014-03-18 Richard Biener <rguenther@suse.de>
178
179 * doc/invoke.texi (flto): Update for changes in 4.9.
180
181 2014-03-18 Richard Biener <rguenther@suse.de>
182
183 * doc/loop.texi: Remove section on the removed lambda framework.
184 Update loop docs with recent changes in preserving loop structure.
185
186 2014-03-18 Richard Biener <rguenther@suse.de>
187
188 * doc/lto.texi (-fresolution): Document.
189
190 2014-03-18 Richard Biener <rguenther@suse.de>
191
192 * doc/contrib.texi: Adjust my name.
193
194 2014-03-18 Jakub Jelinek <jakub@redhat.com>
195
196 PR ipa/58721
197 * internal-fn.c: Include diagnostic-core.h.
198 (expand_BUILTIN_EXPECT): New function.
199 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
200 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
201 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
202 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
203 IFN_BUILTIN_EXPECT.
204 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
205 Revert 3 argument __builtin_expect code.
206 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
207 * gimple-fold.c (gimple_fold_call): Likewise.
208 * tree.h (fold_builtin_expect): New prototype.
209 * builtins.c (build_builtin_expect_predicate): Add predictor
210 argument, if non-NULL, create 3 argument __builtin_expect.
211 (fold_builtin_expect): No longer static. Add ARG2 argument,
212 pass it through to build_builtin_expect_predicate.
213 (fold_builtin_2): Adjust caller.
214 (fold_builtin_3): Handle BUILT_IN_EXPECT.
215 * internal-fn.def (BUILTIN_EXPECT): New.
216
217 2014-03-18 Tobias Burnus <burnus@net-b.de>
218
219 PR ipa/58721
220 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
221 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
222 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
223
224 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
225
226 PR ipa/58721
227 * predict.c (combine_predictions_for_bb): Fix up formatting.
228 (expr_expected_value_1, expr_expected_value): Add predictor argument,
229 fill what it points to if non-NULL.
230 (tree_predict_by_opcode): Adjust caller, use the predictor.
231 * predict.def (PRED_COMPARE_AND_SWAP): Add.
232
233 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
234
235 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
236 proper constant for the store mode.
237
238 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
239
240 * symtab.c (change_decl_assembler_name): Fix transparent alias
241 chain construction.
242
243 2014-03-16 Renlin Li <Renlin.Li@arm.com>
244
245 * config/aarch64/aarch64.c: Correct the comments about the
246 aarch64 stack layout.
247
248 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
249
250 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
251 check for GF_OMP_FOR_KIND_FOR.
252
253 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
254
255 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
256 ymm and zmm register names.
257
258 2014-03-17 Jakub Jelinek <jakub@redhat.com>
259
260 PR target/60516
261 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
262 note creation for the 2010-08-31 changes.
263
264 2014-03-17 Marek Polacek <polacek@redhat.com>
265
266 PR middle-end/60534
267 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
268 as -fno-tree-loop-vectorize.
269 (expand_omp_simd): Likewise.
270
271 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
272
273 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
274 (eligible_for_call_delay): New prototype.
275 * config/sparc/sparc.c (tls_call_delay): Rename into...
276 (eligible_for_call_delay): ...this. Return false if the instruction
277 cannot be put in the delay slot of a branch.
278 (eligible_for_restore_insn): Simplify.
279 (eligible_for_return_delay): Return false if the instruction cannot be
280 put in the delay slot of a branch and simplify.
281 (eligible_for_sibcall_delay): Return false if the instruction cannot be
282 put in the delay slot of a branch.
283 * config/sparc/sparc.md (fix_ut699): New attribute.
284 (tls_call_delay): Delete.
285 (in_call_delay): Reimplement.
286 (eligible_for_sibcall_delay): Rename into...
287 (in_sibcall_delay): ...this.
288 (eligible_for_return_delay): Rename into...
289 (in_return_delay): ...this.
290 (in_branch_delay): Reimplement.
291 (in_uncond_branch_delay): Delete.
292 (in_annul_branch_delay): Delete.
293
294 2014-03-14 Richard Henderson <rth@redhat.com>
295
296 PR target/60525
297 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
298 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
299 (*floathi<X87MODEF>2_i387_with_temp): Remove.
300 (floathi splitters): Remove.
301 (float<SWI48x>xf2): New pattern.
302 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
303 code that tried to handle DImode for 32-bit, but which was excluded
304 by the pattern's condition. Drop allocation of stack temporary.
305 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
306 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
307 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
308 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
309 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
310 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
311 (*float<SWI48><MODEF>2_sse_interunit): Remove.
312 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
313 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
314 (*float<SWI48x><X87MODEF>2_i387): Remove.
315 (all float _with_temp splitters): Remove.
316 (*float<SWI48x><MODEF>2_i387): New pattern.
317 (*float<SWI48><MODEF>2_sse): New pattern.
318 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
319 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
320
321 2014-03-14 Jakub Jelinek <jakub@redhat.com>
322 Marek Polacek <polacek@redhat.com>
323
324 PR middle-end/60484
325 * common.opt (dump_base_name_prefixed): New Variable.
326 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
327 if x_dump_base_name_prefixed is already set, set it at the end.
328
329 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
330
331 PR rtl-optimization/60508
332 * lra-constraints.c (get_reload_reg): Add new parameter
333 in_subreg_p.
334 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
335 Pass the new parameter values.
336
337 2014-03-14 Richard Biener <rguenther@suse.de>
338
339 * common.opt: Revert unintented changes from r205065.
340 * opts.c: Likewise.
341
342 2014-03-14 Richard Biener <rguenther@suse.de>
343
344 PR middle-end/60518
345 * cfghooks.c (split_block): Properly adjust all loops the
346 block was a latch of.
347
348 2014-03-14 Martin Jambor <mjambor@suse.cz>
349
350 PR lto/60461
351 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
352 and simplify it.
353
354 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
355
356 PR target/59396
357 * config/avr/avr.c (avr_set_current_function): Pass function name
358 through default_strip_name_encoding before sanity checking instead
359 of skipping the first char of the assembler name.
360
361 2014-03-13 Richard Henderson <rth@redhat.com>
362
363 PR debug/60438
364 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
365 (ix86_force_to_memory, ix86_free_from_memory): Remove.
366 * config/i386/i386-protos.h: Likewise.
367 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
368 in the expander instead of a splitter.
369 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
370 any possibility of requiring a memory.
371 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
372 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
373 (fp branch splitters): Update for ix86_split_fp_branch.
374 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
375 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
376 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
377 (*fop_<MODEF>_2_i387): Remove f/r alternative.
378 (*fop_<MODEF>_3_i387): Likewise.
379 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
380 (splitters for the fop_* register patterns): Remove.
381 (fscalexf4_i387): Rename from *fscalexf4_i387.
382 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
383
384 2014-03-13 Jakub Jelinek <jakub@redhat.com>
385
386 PR tree-optimization/59779
387 * tree-dfa.c (get_ref_base_and_extent): Use double_int
388 type for bitsize and maxsize instead of HOST_WIDE_INT.
389
390 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
391
392 PR rtl-optimization/57320
393 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
394 the CFG after thread_prologue_and_epilogue_insns.
395
396 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
397
398 PR rtl-optimization/57189
399 * lra-constraints.c (process_alt_operands): Disfavor spilling
400 vector pseudos.
401
402 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
403
404 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic
405 messages.
406
407 2014-03-13 Jakub Jelinek <jakub@redhat.com>
408
409 PR tree-optimization/59025
410 PR middle-end/60418
411 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
412 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
413
414 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
415
416 PR target/60486
417 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
418 calls of avr_out_plus_1.
419
420 2014-03-13 Bin Cheng <bin.cheng@arm.com>
421
422 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
423 BB's single pred and update the father loop's latch info later.
424
425 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
426
427 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
428 (VEC_M): Likewise.
429 (VEC_N): Likewise.
430 (VEC_R): Likewise.
431 (VEC_base): Likewise.
432 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
433 registers, we need to swap double words in little endian mode.
434
435 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
436 to be a container mode for 128-bit integer operations added in ISA
437 2.07. Unlike TImode and PTImode, the preferred register set is
438 the Altivec/VMX registers for the 128-bit operations.
439
440 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
441 declarations.
442 (rs6000_split_128bit_ok_p): Likewise.
443
444 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
445 macros for creating ISA 2.07 normal and overloaded builtin
446 functions with 3 arguments.
447 (BU_P8V_OVERLOAD_3): Likewise.
448 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
449 for use as overloaded functions.
450 (VPERM_1TI_UNS): Likewise.
451 (VSEL_1TI): Likewise.
452 (VSEL_1TI_UNS): Likewise.
453 (ST_INTERNAL_1ti): Likewise.
454 (LD_INTERNAL_1ti): Likewise.
455 (XXSEL_1TI): Likewise.
456 (XXSEL_1TI_UNS): Likewise.
457 (VPERM_1TI): Likewise.
458 (VPERM_1TI_UNS): Likewise.
459 (XXPERMDI_1TI): Likewise.
460 (SET_1TI): Likewise.
461 (LXVD2X_V1TI): Likewise.
462 (STXVD2X_V1TI): Likewise.
463 (VEC_INIT_V1TI): Likewise.
464 (VEC_SET_V1TI): Likewise.
465 (VEC_EXT_V1TI): Likewise.
466 (EQV_V1TI): Likewise.
467 (NAND_V1TI): Likewise.
468 (ORC_V1TI): Likewise.
469 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
470 added in ISA 2.07. Add both normal 'altivec' builtins, and the
471 overloaded builtin.
472 (VADDUQM): Likewise.
473 (VSUBCUQ): Likewise.
474 (VADDEUQM): Likewise.
475 (VADDECUQ): Likewise.
476 (VSUBEUQM): Likewise.
477 (VSUBECUQ): Likewise.
478
479 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
480 __int128_t and __uint128_t types.
481 (__uint128_type): Likewise.
482 (altivec_categorize_keyword): Add support for vector __int128_t,
483 vector __uint128_t, vector __int128, and vector unsigned __int128
484 as a container type for TImode operations that need to be done in
485 VSX/Altivec registers.
486 (rs6000_macro_to_expand): Likewise.
487 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
488 to support 128-bit integer instructions vaddcuq, vadduqm,
489 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
490 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
491
492 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
493 for V1TImode, and set up preferences to use VSX/Altivec registers.
494 Setup VSX reload handlers.
495 (rs6000_debug_reg_global): Likewise.
496 (rs6000_init_hard_regno_mode_ok): Likewise.
497 (rs6000_preferred_simd_mode): Likewise.
498 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
499 (easy_altivec_constant): Likewise.
500 (output_vec_const_move): Likewise.
501 (rs6000_expand_vector_set): Convert V1TImode set and extract to
502 simple move.
503 (rs6000_expand_vector_extract): Likewise.
504 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
505 addressing.
506 (rs6000_const_vec): Add support for V1TImode.
507 (rs6000_emit_le_vsx_load): Swap double words when loading or
508 storing TImode/V1TImode.
509 (rs6000_emit_le_vsx_store): Likewise.
510 (rs6000_emit_le_vsx_move): Likewise.
511 (rs6000_emit_move): Add support for V1TImode.
512 (altivec_expand_ld_builtin): Likewise.
513 (altivec_expand_st_builtin): Likewise.
514 (altivec_expand_vec_init_builtin): Likewise.
515 (altivec_expand_builtin): Likewise.
516 (rs6000_init_builtins): Add support for V1TImode type. Add
517 support for ISA 2.07 128-bit integer builtins. Define type names
518 for the VSX/Altivec vector types.
519 (altivec_init_builtins): Add support for overloaded vector
520 functions with V1TImode type.
521 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
522 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
523 external function.
524 (rs6000_split_128bit_ok_p): Likewise.
525 (rs6000_handle_altivec_attribute): Create V1TImode from vector
526 __int128_t and vector __uint128_t.
527
528 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
529 and mode attributes.
530 (VSX_M): Likewise.
531 (VSX_M2): Likewise.
532 (VSm): Likewise.
533 (VSs): Likewise.
534 (VSr): Likewise.
535 (VSv): Likewise.
536 (VS_scalar): Likewise.
537 (VS_double): Likewise.
538 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
539
540 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
541 we support the ISA 2.07 128-bit integer arithmetic instructions.
542 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
543 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
544 and TImode types for use with the builtin functions.
545 (V1TI_type_node): Likewise.
546 (unsigned_V1TI_type_node): Likewise.
547 (intTI_type_internal_node): Likewise.
548 (uintTI_type_internal_node): Likewise.
549
550 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
551 128-bit builtin functions.
552 (UNSPEC_VADDEUQM): Likewise.
553 (UNSPEC_VADDECUQ): Likewise.
554 (UNSPEC_VSUBCUQ): Likewise.
555 (UNSPEC_VSUBEUQM): Likewise.
556 (UNSPEC_VSUBECUQ): Likewise.
557 (VM): Add V1TImode to vector mode iterators.
558 (VM2): Likewise.
559 (VI_unit): Likewise.
560 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
561 (altivec_vaddcuq): Likewise.
562 (altivec_vsubuqm): Likewise.
563 (altivec_vsubcuq): Likewise.
564 (altivec_vaddeuqm): Likewise.
565 (altivec_vaddecuq): Likewise.
566 (altivec_vsubeuqm): Likewise.
567 (altivec_vsubecuq): Likewise.
568
569 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
570 mode iterators.
571 (BOOL_128): Likewise.
572 (BOOL_REGS_OUTPUT): Likewise.
573 (BOOL_REGS_OP1): Likewise.
574 (BOOL_REGS_OP2): Likewise.
575 (BOOL_REGS_UNARY): Likewise.
576 (BOOL_REGS_AND_CR0): Likewise.
577
578 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
579 128-bit integer builtin support.
580 (vec_vadduqm): Likewise.
581 (vec_vaddecuq): Likewise.
582 (vec_vaddeuqm): Likewise.
583 (vec_vsubecuq): Likewise.
584 (vec_vsubeuqm): Likewise.
585 (vec_vsubcuq): Likewise.
586 (vec_vsubuqm): Likewise.
587
588 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
589 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
590 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
591 128-bit integer add/subtract to ISA 2.07.
592
593 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
594
595 * config/arc/arc.c (arc_predicate_delay_insns):
596 Fix third argument passed to conditionalize_nonjump.
597
598 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
599
600 * config/aarch64/aarch64-builtins.c
601 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
602 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
603 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
604 instead of __builtin_lfloor.
605 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
606
607 2014-03-12 Jakub Jelinek <jakub@redhat.com>
608
609 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
610 (tree_ssa_ifcombine_bb_1): New function.
611 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
612 is an empty forwarder block to then_bb or vice versa and then_bb
613 and else_bb are effectively swapped.
614
615 2014-03-12 Christian Bruel <christian.bruel@st.com>
616
617 PR target/60264
618 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
619 REG_CFA_DEF_CFA note.
620 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
621 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
622
623 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
624
625 PR tree-optimization/60454
626 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
627
628 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
629
630 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
631 Do not define target_cpu_default2 to generic.
632 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
633 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
634 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
635
636 2014-03-12 Jakub Jelinek <jakub@redhat.com>
637 Marc Glisse <marc.glisse@inria.fr>
638
639 PR tree-optimization/60502
640 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
641 instead of build_low_bits_mask.
642
643 2014-03-12 Jakub Jelinek <jakub@redhat.com>
644
645 PR middle-end/60482
646 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
647 if there are multiple uses, but op doesn't live on E edge.
648 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
649 clobber stmts before __builtin_unreachable.
650
651 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
652
653 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
654 hard_frame_pointer_rtx.
655 * cse.c (cse_insn): Remove volatile check.
656 * cselib.c (cselib_process_insn): Likewise.
657 * dse.c (scan_insn): Likewise.
658
659 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
660
661 * config/arc/arc.c (conditionalize_nonjump): New function,
662 broken out of ...
663 (arc_ifcvt): ... this.
664 (arc_predicate_delay_insns): Use it.
665
666 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
667
668 * config/arc/predicates.md (extend_operand): During/after reload,
669 allow const_int_operand.
670 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
671 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
672 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
673 to "i".
674 (umulsi3_highpart_i): Likewise.
675
676 2014-03-11 Richard Biener <rguenther@suse.de>
677
678 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
679 Add asserts to guard possible wrong-code bugs.
680
681 2014-03-11 Richard Biener <rguenther@suse.de>
682
683 PR tree-optimization/60429
684 PR tree-optimization/60485
685 * tree-ssa-structalias.c (set_union_with_increment): Properly
686 take into account all fields that overlap the shifted vars.
687 (do_sd_constraint): Likewise.
688 (do_ds_constraint): Likewise.
689 (get_constraint_for_ptr_offset): Likewise.
690
691 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
692
693 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
694 (nios2_compute_frame_layout):
695 Add calculation of cfun->machine->fp_save_offset.
696 (nios2_expand_prologue): Correct setting of frame pointer register
697 in prologue.
698 (nios2_expand_epilogue): Update recovery of stack pointer from
699 frame pointer accordingly.
700 (nios2_initial_elimination_offset): Update calculation of offset
701 for eliminating to HARD_FRAME_POINTER_REGNUM.
702
703 2014-03-10 Jakub Jelinek <jakub@redhat.com>
704
705 PR ipa/60457
706 * ipa.c (symtab_remove_unreachable_nodes): Don't call
707 cgraph_get_create_node on VAR_DECLs.
708
709 2014-03-10 Richard Biener <rguenther@suse.de>
710
711 PR middle-end/60474
712 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
713
714 2014-03-08 Douglas B Rupp <rupp@gnat.com>
715
716 * config/vms/vms.opt (vms_float_format): New variable.
717
718 2014-03-08 Tobias Burnus <burnus@net-b.de>
719
720 * doc/invoke.texi (-fcilkplus): Update implementation status.
721
722 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
723 Richard Biener <rguenther@suse.de>
724
725 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
726 consistently accross all TUs.
727 (run_gcc): Enable -fshort-double automatically at link at link-time
728 and disallow override.
729
730 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
731
732 PR target/58271
733 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
734 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
735 if they can't be used.
736
737 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
738
739 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
740 for Solaris 11/x86 ld.
741 * configure: Regenerate.
742
743 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
744
745 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
746 (LIB_TLS_SPEC): Save as ld_tls_libs.
747 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
748 (HAVE_AS_IX86_TLSLDM): New test.
749 * configure, config.in: Regenerate.
750 * config/i386/i386.c (legitimize_tls_address): Fall back to
751 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
752 cannot support TLS_MODEL_LOCAL_DYNAMIC.
753 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
754 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
755
756 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
757
758 * common.opt (fira-loop-pressure): Mark as optimization.
759
760 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
761
762 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
763 an OpenMP mappable type.
764
765 2014-03-06 Matthias Klose <doko@ubuntu.com>
766
767 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
768 MULTILIB_OSDIRNAMES is not defined.
769
770 2014-03-06 Jakub Jelinek <jakub@redhat.com>
771 Meador Inge <meadori@codesourcery.com>
772
773 PR target/58595
774 * config/arm/arm.c (arm_tls_symbol_p): Remove.
775 (arm_legitimize_address): Call legitimize_tls_address for any
776 arm_tls_referenced_p expression, handle constant addend. Call it
777 before testing for !TARGET_ARM.
778 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
779
780 2014-03-06 Richard Biener <rguenther@suse.de>
781
782 PR middle-end/60445
783 PR lto/60424
784 PR lto/60427
785 Revert
786 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
787
788 * tree-streamer.c (record_common_node): Assert we don't record
789 nodes with type double.
790 (preload_common_node): Skip type double, complex double and double
791 pointer since it is now frontend dependent due to fshort-double option.
792
793 2014-03-06 Richard Biener <rguenther@suse.de>
794
795 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
796 or -fno-lto is specified and the linker has full plugin support.
797 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
798 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
799 * lto-wrapper.c (merge_and_complain): Merge compile-time
800 optimization levels.
801 (run_gcc): And pass it through to the link options.
802
803 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
804
805 PR debug/60381
806 Revert:
807 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
808 PR debug/59992
809 * cselib.c (remove_useless_values): Skip to avoid quadratic
810 behavior if the condition moved from...
811 (cselib_process_insn): ... here holds.
812
813 2014-03-05 Jakub Jelinek <jakub@redhat.com>
814
815 PR plugins/59335
816 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
817 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
818
819 PR plugins/59335
820 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
821 (TM_H): Add x86-tune.def.
822
823 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
824
825 * config/aarch64/aarch64.c (generic_tunings):
826 Use cortexa57_extra_costs.
827
828 2014-03-05 Jakub Jelinek <jakub@redhat.com>
829
830 PR lto/60404
831 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
832 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
833 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
834 cost for in_lto_p.
835
836 2014-03-04 Heiher <r@hev.cc>
837
838 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
839 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
840
841 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
842
843 * config/i386/predicates.md (const2356_operand): Change to ...
844 (const2367_operand): ... this.
845 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
846 const2367_operand.
847 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
848 (*avx512pf_scatterpf<mode>sf): Ditto.
849 (avx512pf_scatterpf<mode>df): Ditto.
850 (*avx512pf_scatterpf<mode>df_mask): Ditto.
851 (*avx512pf_scatterpf<mode>df): Ditto.
852 * config/i386/i386.c (ix86_expand_builtin): Update
853 incorrect hint operand error message.
854
855 2014-03-04 Richard Biener <rguenther@suse.de>
856
857 * lto-section-in.c (lto_get_section_data): Fix const cast.
858
859 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
860
861 * tree-streamer.c (record_common_node): Assert we don't record
862 nodes with type double.
863 (preload_common_node): Skip type double, complex double and double
864 pointer since it is now frontend dependent due to fshort-double option.
865
866 2014-03-04 Richard Biener <rguenther@suse.de>
867
868 PR lto/60405
869 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
870 (lto_input_toplevel_asms): Likewise.
871 * lto-section-in.c (lto_get_section_data): Instead do it here
872 for every section.
873
874 2014-03-04 Richard Biener <rguenther@suse.de>
875
876 PR tree-optimization/60382
877 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
878 dead PHIs a reduction.
879
880 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
881
882 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
883 hint value.
884 (_mm_prefetch): Move out of GCC target("sse") pragma.
885 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
886 GCC target("prfchw") pragma.
887 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
888 for locality <= 2.
889 * config/i386/i386.c (ix86_option_override_internal): Enable
890 -mprfchw with -mprefetchwt1.
891
892 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
893
894 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
895 Mark as varying.
896
897 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
898
899 * opts.h (CL_PCH_IGNORE): Define.
900 * targhooks.c (option_affects_pch_p):
901 Return false for options that have CL_PCH_IGNORE set.
902 * opt-functions.awk: Process PchIgnore.
903 * doc/options.texi: Document PchIgnore.
904
905 * config/arc/arc.opt (misize): Add PchIgnore property.
906
907 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
908
909 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
910 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
911 constraint on constants to permit them being loaded into
912 GENERAL_REGS or BASE_REGS.
913
914 2014-03-03 Nick Clifton <nickc@redhat.com>
915
916 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
917 anti-cacnonical alternatives.
918 (negandhi3_real): New pattern.
919 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
920
921 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
922
923 * config/avr/avr-mcus.def: Remove atxmega16x1.
924 * config/avr/avr-tables.opt: Regenerate.
925 * config/avr/t-multilib: Regenerate.
926 * doc/avr-mmcu.texi: Regenerate.
927
928 2014-03-03 Tobias Grosser <tobias@grosser.es>
929 Mircea Namolaru <mircea.namolaru@inria.fr>
930
931 PR tree-optimization/58028
932 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
933 scalar dimensions.
934
935 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
936
937 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
938 not handled by recognizers.
939
940 2014-03-03 Jakub Jelinek <jakub@redhat.com>
941
942 PR middle-end/60175
943 * function.c (expand_function_end): Don't emit
944 clobber_return_register sequence if clobber_after is a BARRIER.
945 * cfgexpand.c (construct_exit_block): Append instructions before
946 return_label to prev_bb.
947
948 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
949
950 * config/rs6000/constraints.md: Document reserved use of "wc".
951
952 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
953
954 PR ipa/60150
955 * ipa.c (function_and_variable_visibility): When dissolving comdat
956 group, also set all symbols to local.
957
958 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
959
960 PR ipa/60306
961
962 Revert:
963 2013-12-14 Jan Hubicka <jh@suse.cz>
964 PR middle-end/58477
965 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
966
967 2014-03-02 Jon Beniston <jon@beniston.com>
968
969 PR bootstrap/48230
970 PR bootstrap/50927
971 PR bootstrap/52466
972 PR target/46898
973 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
974 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
975 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
976 (simple_return, *simple_return): New patterns
977 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
978 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
979
980 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
981
982 * dwarf2out.c (gen_subprogram_die): Tidy.
983
984 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
985
986 PR target/60071
987 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
988 (*mov_t_msb_neg_negc): ... this new insn.
989
990 2014-02-28 Jason Merrill <jason@redhat.com>
991
992 PR c++/58678
993 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
994 function.
995
996 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
997
998 PR c++/60314
999 * dwarf2out.c (decltype_auto_die): New static.
1000 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
1001 (gen_type_die_with_usage): Handle 'decltype(auto)'.
1002 (is_cxx_auto): Likewise.
1003
1004 2014-02-28 Ian Bolton <ian.bolton@arm.com>
1005
1006 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
1007 we are not using general regs only.
1008
1009 2014-02-28 Richard Biener <rguenther@suse.de>
1010
1011 PR target/60280
1012 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
1013 previous fix and only allow to remove trivial pre-headers
1014 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
1015 (remove_forwarder_block): Properly update the latch of a loop.
1016
1017 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1018
1019 PR debug/59992
1020 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
1021 (cselib_preserved_hash_table): New.
1022 (preserve_constants_and_equivs): Move preserved vals to it.
1023 (cselib_find_slot): Look it up first.
1024 (cselib_init): Initialize it.
1025 (cselib_finish): Release it.
1026 (dump_cselib_table): Dump it.
1027
1028 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1029
1030 PR debug/59992
1031 * cselib.c (remove_useless_values): Skip to avoid quadratic
1032 behavior if the condition moved from...
1033 (cselib_process_insn): ... here holds.
1034
1035 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1036
1037 PR debug/57232
1038 * var-tracking.c (vt_initialize): Apply the same condition to
1039 preserve the CFA base value.
1040
1041 2014-02-28 Joey Ye <joey.ye@arm.com>
1042
1043 PR target/PR60169
1044 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
1045 if reload in progress or completed.
1046
1047 2014-02-28 Tobias Burnus <burnus@net-b.de>
1048
1049 PR middle-end/60147
1050 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
1051 NAMELIST_DECL.
1052
1053 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
1054
1055 * doc/tm.texi.in (Condition Code Status): Update documention for
1056 relative locations of cc0-setter and cc0-user.
1057
1058 2014-02-27 Jeff Law <law@redhat.com>
1059
1060 PR rtl-optimization/52714
1061 * combine.c (try_combine): When splitting an unrecognized PARALLEL
1062 into two independent simple sets, if I3 is a jump, ensure the
1063 pattern we place into I3 is a (set (pc) ...).
1064
1065 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
1066 Jeff Law <law@redhat.com>
1067
1068 PR rtl-optimization/49847
1069 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
1070 are in different blocks.
1071 * doc/tm.texi (Condition Code Status): Update documention for
1072 relative locations of cc0-setter and cc0-user.
1073
1074 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
1075
1076 PR target/59222
1077 * lra.c (lra_emit_add): Check SUBREG too.
1078
1079 2014-02-27 Andreas Schwab <schwab@suse.de>
1080
1081 * config/m68k/m68k.c (m68k_option_override): Disable
1082 -flive-range-shrinkage for classic m68k.
1083 (m68k_override_options_after_change): Likewise.
1084
1085 2014-02-27 Marek Polacek <polacek@redhat.com>
1086
1087 PR middle-end/59223
1088 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
1089 -Wmaybe-uninitialized.
1090
1091 2014-02-27 Alan Modra <amodra@gmail.com>
1092
1093 PR target/57936
1094 * reload1.c (emit_input_reload_insns): When reload_override_in,
1095 set old to rl->in_reg when rl->in_reg is a subreg.
1096
1097 2014-02-26 Richard Biener <rguenther@suse.de>
1098
1099 PR bootstrap/60343
1100 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
1101
1102 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1103
1104 * common/config/i386/predicates.md (const1256_operand): Remove.
1105 (const2356_operand): New.
1106 (const_1_to_2_operand): Remove.
1107 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1108 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1109 (*avx512pf_gatherpf<mode>sf): Ditto.
1110 (avx512pf_gatherpf<mode>df): Ditto.
1111 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1112 (*avx512pf_gatherpf<mode>df): Ditto.
1113 (avx512pf_scatterpf<mode>sf): Ditto.
1114 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1115 (*avx512pf_scatterpf<mode>sf): Ditto.
1116 (avx512pf_scatterpf<mode>df): Ditto.
1117 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1118 (*avx512pf_scatterpf<mode>df): Ditto.
1119 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
1120
1121 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
1122
1123 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
1124 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
1125 (_mm512_mask_testn_epi64_mask): Move to ...
1126 * config/i386/avx512cdintrin.h: Here.
1127 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
1128 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
1129 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
1130 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
1131 TARGET_AVX512F from TARGET_AVX512CD.
1132
1133 2014-02-26 Richard Biener <rguenther@suse.de>
1134
1135 PR ipa/60327
1136 * ipa.c (walk_polymorphic_call_targets): Properly guard
1137 call to inline_update_overall_summary.
1138
1139 2014-02-26 Bin Cheng <bin.cheng@arm.com>
1140
1141 PR target/60280
1142 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
1143 and latches only if requested. Fix latch if it is removed.
1144 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
1145 LOOPS_HAVE_PREHEADERS.
1146
1147 2014-02-25 Andrew Pinski <apinski@cavium.com>
1148
1149 * builtins.c (expand_builtin_thread_pointer): Create a new target
1150 when the target is NULL.
1151
1152 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
1153
1154 PR rtl-optimization/60317
1155 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1156 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1157 * lra-assigns.c: Include params.h.
1158 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
1159 other reload pseudos considerations.
1160
1161 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1162
1163 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
1164 to use canonical form for nor<mode>3.
1165
1166 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1167
1168 PR target/55426
1169 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
1170 conversions.
1171
1172 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1173
1174 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
1175 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
1176 (ix86_handle_option): Handle OPT_mprefetchwt1.
1177 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
1178 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
1179 PREFETCHWT1 CPUID.
1180 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1181 OPTION_MASK_ISA_PREFETCHWT1.
1182 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
1183 (PTA_PREFETCHWT1): New.
1184 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
1185 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
1186 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
1187 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
1188 (*prefetch_avx512pf_<mode>_: Change into ...
1189 (*prefetch_prefetchwt1_<mode>: This.
1190 * config/i386/i386.opt (mprefetchwt1): New.
1191 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
1192 (_mm_prefetch): Handle intent to write.
1193 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
1194
1195 2014-02-25 Richard Biener <rguenther@suse.de>
1196
1197 PR middle-end/60291
1198 * emit-rtl.c (mem_attrs_htab): Remove.
1199 (mem_attrs_htab_hash): Likewise.
1200 (mem_attrs_htab_eq): Likewise.
1201 (set_mem_attrs): Always allocate new mem-attrs when something changed.
1202 (init_emit_once): Do not allocate mem_attrs_htab.
1203
1204 2014-02-25 Richard Biener <rguenther@suse.de>
1205
1206 PR lto/60319
1207 * lto-opts.c (lto_write_options): Output non-explicit conservative
1208 -fwrapv, -fno-trapv and -fno-strict-overflow.
1209 * lto-wrapper.c (merge_and_complain): Handle merging those options.
1210 (run_gcc): And pass them through.
1211
1212 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1213
1214 * sel-sched.c (calculate_new_fences): New parameter ptime.
1215 Calculate it as a maximum over all fence cycles.
1216 (sel_sched_region_2): Adjust the call to calculate_new_fences.
1217 Print the final schedule timing when sched_verbose.
1218
1219 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1220
1221 PR rtl-optimization/60292
1222 * sel-sched.c (fill_vec_av_set): Do not reset target availability
1223 bit fot the fence instruction.
1224
1225 2014-02-24 Alangi Derick <alangiderick@gmail.com>
1226
1227 * calls.h: Fix typo in comment.
1228
1229 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
1230
1231 * config/pa/pa.c (pa_output_move_double): Don't valididate when
1232 adjusting offsetable addresses.
1233
1234 2014-02-24 Guozhi Wei <carrot@google.com>
1235
1236 * sparseset.h (sparseset_pop): Fix the wrong index.
1237
1238 2014-02-24 Walter Lee <walt@tilera.com>
1239
1240 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
1241 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
1242 triplet.
1243 * common/config/tilegx/tilegx-common.c
1244 (TARGET_DEFAULT_TARGET_FLAGS): Define.
1245 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
1246 (LINK_SPEC): Ditto.
1247 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
1248 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
1249 (tilegx_gimplify_va_arg_expr): Handle big endian.
1250 (tilegx_expand_unaligned_load): Ditto.
1251 (tilegx_expand_unaligned_store): Ditto.
1252 (TARGET_RETURN_IN_MSB): New.
1253 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
1254 (TARGET_ENDIAN_DEFAULT): New.
1255 (TARGET_BIG_ENDIAN): Handle big endian.
1256 (BYTES_BIG_ENDIAN): Ditto.
1257 (WORDS_BIG_ENDIAN): Ditto.
1258 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
1259 (ENDIAN_SPEC): New.
1260 (EXTRA_SPECS): New.
1261 * config/tilegx/tilegx.md (extv): Handle big endian.
1262 (extzv): Ditto.
1263 (insn_st<n>): Ditto.
1264 (insn_st<n>_add<bitsuffix>): Ditto.
1265 (insn_stnt<n>): Ditto.
1266 (insn_stnt<n>_add<bitsuffix>):Ditto.
1267 (vec_interleave_highv8qi): Handle big endian.
1268 (vec_interleave_highv8qi_be): New.
1269 (vec_interleave_highv8qi_le): New.
1270 (insn_v1int_h): Handle big endian.
1271 (vec_interleave_lowv8qi): Handle big endian.
1272 (vec_interleave_lowv8qi_be): New.
1273 (vec_interleave_lowv8qi_le): New.
1274 (insn_v1int_l): Handle big endian.
1275 (vec_interleave_highv4hi): Handle big endian.
1276 (vec_interleave_highv4hi_be): New.
1277 (vec_interleave_highv4hi_le): New.
1278 (insn_v2int_h): Handle big endian.
1279 (vec_interleave_lowv4hi): Handle big endian.
1280 (vec_interleave_lowv4hi_be): New.
1281 (vec_interleave_lowv4hi_le): New.
1282 (insn_v2int_l): Handle big endian.
1283 (vec_interleave_highv2si): Handle big endian.
1284 (vec_interleave_highv2si_be): New.
1285 (vec_interleave_highv2si_le): New.
1286 (insn_v4int_h): Handle big endian.
1287 (vec_interleave_lowv2si): Handle big endian.
1288 (vec_interleave_lowv2si_be): New.
1289 (vec_interleave_lowv2si_le): New.
1290 (insn_v4int_l): Handle big endian.
1291 * config/tilegx/tilegx.opt (mbig-endian): New option.
1292 (mlittle-endian): New option.
1293 * doc/install.texi: Document tilegxbe-linux.
1294 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
1295
1296 2014-02-24 Martin Jambor <mjambor@suse.cz>
1297
1298 PR ipa/60266
1299 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
1300 there are no parameter descriptors.
1301
1302 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
1303
1304 PR rtl-optimization/60268
1305 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
1306 initialization to ...
1307 (sched_rgn_init): ... here.
1308 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
1309
1310 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1311
1312 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
1313 names.
1314
1315 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
1316
1317 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
1318 definition.
1319
1320 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1321
1322 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
1323 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
1324
1325 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1326
1327 * config/microblaze/predicates.md: Add cmp_op predicate.
1328 * config/microblaze/microblaze.md: Add branch_compare instruction
1329 which uses cmp_op predicate and emits cmp insn before branch.
1330 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
1331 to microblaze_expand_conditional_branch and consolidate logic.
1332 (microblaze_expand_conditional_branch): emit branch_compare
1333 insn instead of handling cmp op separate from branch insn.
1334
1335 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1336
1337 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
1338 to permit subregs.
1339
1340 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1341
1342 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
1343 define_insn with define_expand and new define_insn
1344 *altivec_lve<VI_char>x_internal.
1345 (altivec_stve<VI_char>x): Replace define_insn with define_expand
1346 and new define_insn *altivec_stve<VI_char>x_internal.
1347 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
1348 prototype.
1349 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
1350 lve*x built-ins.
1351 (altivec_expand_stvex_be): New function.
1352
1353 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
1354
1355 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
1356 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
1357 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
1358 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
1359
1360 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
1361
1362 PR target/60298
1363 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
1364 instead of emit_move_insn.
1365
1366 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1367
1368 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
1369 vspltw with vsldoi.
1370 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
1371 gen_altivec_vsumsws.
1372
1373 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1374
1375 * config/rs6000/altivec.md (altivec_lvxl): Rename as
1376 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
1377 (altivec_lvxl_<mode>): New define_expand incorporating
1378 -maltivec=be semantics where needed.
1379 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
1380 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
1381 semantics where needed.
1382 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
1383 (altivec_stvx_<mode>): New define_expand incorporating
1384 -maltivec=be semantics where needed.
1385 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
1386 VM2 iterator instead of V4SI.
1387 (altivec_stvxl_<mode>): New define_expand incorporating
1388 -maltivec=be semantics where needed.
1389 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
1390 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
1391 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
1392 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
1393 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
1394 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
1395 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
1396 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
1397 ALTIVEC_BUILTIN_STVXL.
1398 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
1399 (altivec_expand_stvx_be): Likewise.
1400 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
1401 (altivec_expand_lvx_be): Likewise.
1402 (altivec_expand_stvx_be): Likewise.
1403 (altivec_expand_builtin): Add cases for
1404 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
1405 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
1406 (altivec_init_builtins): Add definitions for
1407 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
1408 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
1409
1410 2014-02-21 Catherine Moore <clm@codesourcery.com>
1411
1412 * doc/invoke.texi (mvirt, mno-virt): Document.
1413 * config/mips/mips.opt (mvirt): New option.
1414 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
1415
1416 2014-02-21 Richard Biener <rguenther@suse.de>
1417
1418 PR tree-optimization/60276
1419 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
1420 (STMT_VINFO_MIN_NEG_DIST): New macro.
1421 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
1422 STMT_VINFO_MIN_NEG_DIST.
1423 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
1424 made for negative dependence distances still hold.
1425
1426 2014-02-21 Richard Biener <rguenther@suse.de>
1427
1428 PR middle-end/60291
1429 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
1430 DECL_INITIAL for globals not in the current function context.
1431
1432 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1433
1434 PR tree-optimization/56490
1435 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
1436 * tree-ssa-uninit.c: Include params.h.
1437 (compute_control_dep_chain): Add num_calls argument, return false
1438 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
1439 num_calls to recursive call.
1440 (find_predicates): Change dep_chain into normal array,
1441 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
1442 variable and adjust compute_control_dep_chain caller.
1443 (find_def_preds): Likewise.
1444
1445 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
1446
1447 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
1448 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
1449
1450 2014-02-21 Nick Clifton <nickc@redhat.com>
1451
1452 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
1453 (pushhi1): Likewise.
1454 (popqi1): Add mode to pre_dec.
1455 (pophi1): Likewise.
1456
1457 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1458
1459 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
1460 mode for mask of V8SFmode permutation.
1461
1462 2014-02-20 Richard Henderson <rth@redhat.com>
1463
1464 PR c++/60272
1465 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
1466 a new pseudo for OLDVAL.
1467
1468 2014-02-20 Jakub Jelinek <jakub@redhat.com>
1469
1470 PR target/57896
1471 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
1472 gen_reg_rtx if d->testing_p.
1473 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
1474 if d->testing_p and we will certainly return true.
1475 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
1476 if d->testing_p.
1477
1478 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
1479
1480 * emit-rtl.c (gen_reg_rtx): Assert that
1481 crtl->emit.regno_pointer_align_length is non-zero.
1482
1483 2014-02-20 Richard Henderson <rth@redhat.com>
1484
1485 PR c++/60272
1486 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
1487 on failure the store back into EXPECT.
1488
1489 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
1490 Sandra Loosemore <sandra@codesourcery.com>
1491
1492 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
1493 * config/nios2/nios2.c (nios2_function_profiler): Add
1494 -fPIC (flag_pic == 2) support.
1495 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
1496 (nios2_large_offset_p): New function.
1497 (nios2_unspec_reloc_p): Move up position, update to use
1498 nios2_large_offset_p.
1499 (nios2_unspec_address): Remove function.
1500 (nios2_unspec_offset): New function.
1501 (nios2_large_got_address): New function.
1502 (nios2_got_address): Add large offset support.
1503 (nios2_legitimize_tls_address): Update usage of removed and new
1504 functions.
1505 (nios2_symbol_binds_local_p): New function.
1506 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
1507 (nios2_legitimize_address): Update to use nios2_large_offset_p.
1508 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
1509 (nios2_print_operand): Merge H/L processing, add hiadj/lo
1510 processing for (const (unspec ...)).
1511 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
1512
1513 2014-02-20 Richard Biener <rguenther@suse.de>
1514
1515 * tree-cfg.c (replace_uses_by): Mark altered BBs before
1516 doing the substitution.
1517 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
1518
1519 2014-02-20 Martin Jambor <mjambor@suse.cz>
1520
1521 PR ipa/55260
1522 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
1523 info when checking whether lattices are bottom.
1524
1525 2014-02-20 Richard Biener <rguenther@suse.de>
1526
1527 PR middle-end/60221
1528 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
1529 regions at -O0.
1530
1531 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
1532
1533 PR ipa/58555
1534 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
1535 parameter specifying the scaling.
1536 (inline_call): Update.
1537 (want_inline_recursively): Guard division by zero.
1538 (recursive_inlining): Update.
1539 * ipa-inline.h (clone_inlined_nodes): Update.
1540
1541 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1542
1543 PR target/60204
1544 * config/i386/i386.c (classify_argument): Pass structures of size
1545 64 bytes or less in register.
1546
1547 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1548 Kirill Yukhin <kirill.yukhin@intel.com>
1549
1550 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
1551 (_mm_rcp28_round_ss): Ditto.
1552 (_mm_rsqrt28_round_sd): Ditto.
1553 (_mm_rsqrt28_round_ss): Ditto.
1554 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
1555 (_mm_rcp14_round_ss): Ditto.
1556 (_mm_rsqrt14_round_sd): Ditto.
1557 (_mm_rsqrt14_round_ss): Ditto.
1558 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
1559 the first input operand, get rid of match_dup.
1560 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
1561 attribute to sse.
1562 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1563 Ditto.
1564 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
1565 operand as the first input operand, set type attribute.
1566 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1567 Set type attribute.
1568 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
1569 operand as the first input operand, set type attribute.
1570
1571 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1572
1573 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
1574 bit of zero.
1575
1576 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
1577
1578 PR target/60207
1579 * config/i386/i386.c (construct_container): Remove TFmode check
1580 for X86_64_INTEGER_CLASS.
1581
1582 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
1583
1584 PR target/59794
1585 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
1586 only when -Wpsabi is enabled.
1587
1588 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
1589
1590 PR target/59799
1591 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
1592 passing arrays in registers are the same as for structs, so remove the
1593 special case for them.
1594
1595 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
1596
1597 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
1598 destination type, extract only the valid bits if the source type is not
1599 integral and has a different mode.
1600
1601 2014-02-19 Richard Biener <rguenther@suse.de>
1602
1603 PR ipa/60243
1604 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
1605 for all calls.
1606
1607 2014-02-19 Richard Biener <rguenther@suse.de>
1608
1609 PR ipa/60243
1610 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
1611 (ipa_modify_call_arguments): Emit an argument load explicitely and
1612 preserve virtual SSA form there and for the replacement call.
1613 Do not update SSA form nor free dominance info.
1614
1615 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1616
1617 * ipa.c (function_and_variable_visibility): Also clear WEAK
1618 flag when disolving COMDAT_GROUP.
1619
1620 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1621
1622 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
1623 * ipa-prop.c (ipa_set_jf_known_type): Return early when
1624 not devirtualizing.
1625 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
1626 do more sanity checks.
1627 (detect_type_change): Return true when giving up early.
1628 (compute_complex_assign_jump_func): Fix type parameter of
1629 ipa_set_ancestor_jf.
1630 (compute_complex_ancestor_jump_func): Likewise.
1631 (update_jump_functions_after_inlining): Fix updating of
1632 ancestor function.
1633 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
1634
1635 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1636
1637 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
1638 inline clones when edge disappears.
1639
1640 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
1641
1642 PR target/60203
1643 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
1644 Split 64-bit moves into 2 patterns. Do not allow the use of
1645 direct move for TDmode in little endian, since the decimal value
1646 has little endian bytes within a word, but the 64-bit pieces are
1647 ordered in a big endian fashion, and normal subreg's of TDmode are
1648 not allowed.
1649 (mov<mode>_64bit_dm): Likewise.
1650 (movtd_64bit_nodm): Likewise.
1651
1652 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1653
1654 PR tree-optimization/60174
1655 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
1656 statement of an SSA_NAME that occurs in an abnormal PHI node.
1657
1658 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1659
1660 PR sanitizer/60142
1661 * final.c (SEEN_BB): Remove.
1662 (SEEN_NOTE, SEEN_EMITTED): Renumber.
1663 (final_scan_insn): Don't force_source_line on second
1664 NOTE_INSN_BASIC_BLOCK.
1665
1666 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
1667
1668 PR target/60205
1669 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
1670 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
1671 (type_natural_mode): Warn ABI change when %zmm register is not
1672 available for AVX512F vector value passing.
1673
1674 2014-02-18 Kai Tietz <ktietz@redhat.com>
1675
1676 PR target/60193
1677 * config/i386/i386.c (ix86_expand_prologue): Use value in
1678 rax register as displacement when restoring %r10 or %rax.
1679 Fix wrong offset when restoring both registers.
1680
1681 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1682
1683 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
1684 assertion with conditional return.
1685
1686 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1687 Uros Bizjak <ubizjak@gmail.com>
1688
1689 PR driver/60233
1690 * config/i386/driver-i386.c (host_detect_local_cpu): If
1691 YMM state is not saved by the OS, also clear has_f16c. Move
1692 CPUID 0x80000001 handling before YMM state saving checking.
1693
1694 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
1695
1696 PR rtl-optimization/58960
1697 * haifa-sched.c (alloc_global_sched_pressure_data): New,
1698 factored out from ...
1699 (sched_init): ... here.
1700 (free_global_sched_pressure_data): New, factored out from ...
1701 (sched_finish): ... here.
1702 * sched-int.h (free_global_sched_pressure_data): Declare.
1703 * sched-rgn.c (nr_regions_initial): New static global.
1704 (haifa_find_rgns): Initialize it.
1705 (schedule_region): Disable sched-pressure for the newly
1706 generated regions.
1707
1708 2014-02-17 Richard Biener <rguenther@suse.de>
1709
1710 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
1711 release SSA defs of pattern stmts.
1712
1713 2014-02-17 Richard Biener <rguenther@suse.de>
1714
1715 * tree-inline.c (expand_call_inline): Release the virtual
1716 operand defined by the call we are about to inline.
1717
1718 2014-02-17 Richard Biener <rguenther@suse.de>
1719
1720 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
1721
1722 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
1723 Ilya Tocar <ilya.tocar@intel.com>
1724
1725 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
1726 arguments order in builtin.
1727 (_mm512_permutexvar_epi64): Ditto.
1728 (_mm512_mask_permutexvar_epi64): Ditto
1729 (_mm512_maskz_permutexvar_epi32): Ditto
1730 (_mm512_permutexvar_epi32): Ditto
1731 (_mm512_mask_permutexvar_epi32): Ditto
1732
1733 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1734
1735 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
1736 (p8_vmrgow): Likewise.
1737
1738 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1739
1740 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
1741 endian targets.
1742
1743 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1744
1745 PR target/60203
1746 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
1747 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
1748 into 64-bit and 32-bit moves. On 64-bit moves, add support for
1749 using direct move instructions on ISA 2.07. Also adjust
1750 instruction length for 64-bit.
1751 (mov<mode>_64bit, TFmode/TDmode): Likewise.
1752 (mov<mode>_32bit, TFmode/TDmode): Likewise.
1753
1754 2014-02-15 Alan Modra <amodra@gmail.com>
1755
1756 PR target/58675
1757 PR target/57935
1758 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
1759 find_replacement on parts of insn rtl that might be reloaded.
1760
1761 2014-02-15 Richard Biener <rguenther@suse.de>
1762
1763 PR tree-optimization/60183
1764 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
1765 (tree_ssa_phiprop): Calculate and free post-dominators.
1766
1767 2014-02-14 Jeff Law <law@redhat.com>
1768
1769 PR rtl-optimization/60131
1770 * ree.c (get_extended_src_reg): New function.
1771 (combine_reaching_defs): Use it rather than assuming location of REG.
1772 (find_and_remove_re): Verify first operand of extension is
1773 a REG before adding the insns to the copy list.
1774
1775 2014-02-14 Roland McGrath <mcgrathr@google.com>
1776
1777 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
1778 * configure: Regenerated.
1779 * config.in: Regenerated.
1780 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
1781 instead of ASM_SHORT.
1782
1783 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
1784 Richard Earnshaw <rearnsha@arm.com>
1785
1786 PR rtl-optimization/59535
1787 * lra-constraints.c (process_alt_operands): Encourage alternative
1788 when unassigned pseudo class is superset of the alternative class.
1789 (inherit_reload_reg): Don't inherit when optimizing for code size.
1790 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
1791 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
1792 modes not less than 4 for Thumb1.
1793
1794 2014-02-14 Kyle McMartin <kyle@redhat.com>
1795
1796 PR pch/60010
1797 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
1798
1799 2014-02-14 Richard Biener <rguenther@suse.de>
1800
1801 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
1802 (get_frame_arg): Drop the assert with langhook types_compatible_p.
1803 Do not strip INDIRECT_REFs.
1804
1805 2014-02-14 Richard Biener <rguenther@suse.de>
1806
1807 PR lto/60179
1808 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
1809 DECL_FUNCTION_SPECIFIC_TARGET.
1810 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
1811 * tree-streamer-out.c (pack_ts_target_option): Remove.
1812 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
1813 (write_ts_function_decl_tree_pointers): Do not stream
1814 DECL_FUNCTION_SPECIFIC_TARGET.
1815 * tree-streamer-in.c (unpack_ts_target_option): Remove.
1816 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
1817 (lto_input_ts_function_decl_tree_pointers): Do not stream
1818 DECL_FUNCTION_SPECIFIC_TARGET.
1819
1820 2014-02-14 Jakub Jelinek <jakub@redhat.com>
1821
1822 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
1823 (get_initial_def_for_induction, vectorizable_induction): Ignore
1824 debug stmts when looking for exit_phi.
1825 (vectorizable_live_operation): Fix up condition.
1826
1827 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1828
1829 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
1830 nreverse() because it changes the content of original tree list.
1831
1832 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1833
1834 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
1835 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
1836
1837 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1838
1839 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
1840 GNU coding standards.
1841
1842 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1843
1844 PR debug/60152
1845 * dwarf2out.c (gen_subprogram_die): Don't call
1846 add_calling_convention_attribute if subr_die is old_die.
1847
1848 2014-02-13 Sharad Singhai <singhai@google.com>
1849
1850 * doc/optinfo.texi: Fix order of nodes.
1851
1852 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
1853
1854 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
1855 operands[2], not operands[3].
1856
1857 2014-02-13 Richard Biener <rguenther@suse.de>
1858
1859 PR bootstrap/59878
1860 * doc/install.texi (ISL): Update recommended version to 0.12.2,
1861 mention the possibility of an in-tree build.
1862 (CLooG): Update recommended version to 0.18.1, mention the
1863 possibility of an in-tree build and clarify that the ISL
1864 bundled with CLooG does not work.
1865
1866 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1867
1868 PR target/43546
1869 * expr.c (compress_float_constant): If x is a hard register,
1870 extend into a pseudo and then move to x.
1871
1872 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
1873
1874 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
1875 caused by bad second argument to warning_at() with -mhotpatch and
1876 nested functions (e.g. with gfortran).
1877
1878 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
1879
1880 * opts.c (option_name): Remove "enabled by default" rider.
1881
1882 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
1883
1884 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
1885
1886 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
1887 Uros Bizjak <ubizjak@gmail.com>
1888
1889 PR target/60151
1890 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
1891 * configure: Regenerated.
1892
1893 2014-02-12 Richard Biener <rguenther@suse.de>
1894
1895 * vec.c (vec_prefix::calculate_allocation): Move as
1896 inline variant to vec.h.
1897 (vec_prefix::calculate_allocation_1): New out-of-line version.
1898 * vec.h (vec_prefix::calculate_allocation_1): Declare.
1899 (vec_prefix::m_has_auto_buf): Rename to ...
1900 (vec_prefix::m_using_auto_storage): ... this.
1901 (vec_prefix::calculate_allocation): Inline the easy cases
1902 and dispatch to calculate_allocation_1 which doesn't need the
1903 prefix address.
1904 (va_heap::reserve): Use gcc_checking_assert.
1905 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
1906 m_using_auto_storage.
1907 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
1908 member and adjust.
1909 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
1910 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
1911 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
1912
1913 2014-02-12 Richard Biener <rguenther@suse.de>
1914
1915 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
1916 when we found a dependence.
1917
1918 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
1919
1920 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
1921 common code...
1922 (maybe_fold_stmt): ... into this new function.
1923 * omp-low.c (lower_omp): Update comment.
1924
1925 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
1926 last use.
1927
1928 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
1929 dereference.
1930
1931 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
1932
1933 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
1934 identifiers in comments.
1935 (cortexa53_extra_costs): Likewise.
1936 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
1937 (cortexa7_extra_costs): Likewise.
1938 (cortexa12_extra_costs): Likewise.
1939 (cortexa15_extra_costs): Likewise.
1940 (v7m_extra_costs): Likewise.
1941
1942 2014-02-12 Richard Biener <rguenther@suse.de>
1943
1944 PR middle-end/60092
1945 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
1946 of posix_memalign being successful.
1947 (lower_stmt): Restrict lowering of posix_memalign to when
1948 -ftree-bit-ccp is enabled.
1949
1950 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1951
1952 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
1953 arg_loc.
1954 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
1955
1956 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
1957
1958 PR rtl-optimization/60116
1959 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
1960 other_insn once the combination has been validated.
1961
1962 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
1963
1964 PR lto/59468
1965 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
1966 and wrapper.
1967 * ipa-devirt.c: Include demangle.h
1968 (odr_violation_reported): New static variable.
1969 (add_type_duplicate): Update odr_violations.
1970 (maybe_record_node): Add completep parameter; update it.
1971 (record_target_from_binfo): Add COMPLETEP parameter;
1972 update it as needed.
1973 (possible_polymorphic_call_targets_1): Likewise.
1974 (struct polymorphic_call_target_d): Add nonconstruction_targets;
1975 rename FINAL to COMPLETE.
1976 (record_targets_from_bases): Sanity check we found the binfo;
1977 fix COMPLETEP updating.
1978 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
1979 parameter, fix computing of COMPLETEP.
1980 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
1981 at LTO time do demangling.
1982 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
1983 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
1984 parameter.
1985 (gimple_get_virt_method_for_binfo): Likewise.
1986 * gimple-fold.h (gimple_get_virt_method_for_binfo,
1987 gimple_get_virt_method_for_vtable): Update prototypes.
1988
1989 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
1990
1991 PR target/49008
1992 * genautomata.c (add_presence_absence): Fix typo with
1993 {final_}presence_list.
1994
1995 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
1996
1997 PR target/60137
1998 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
1999 for VSX/Altivec vectors that land in GPR registers.
2000
2001 2014-02-11 Richard Henderson <rth@redhat.com>
2002 Jakub Jelinek <jakub@redhat.com>
2003
2004 PR debug/59776
2005 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
2006 around drhs if type conversion to lacc->type is not useless.
2007
2008 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2009
2010 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
2011 tuning struct.
2012 (cortex-a57.cortex-a53): Likewise.
2013 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
2014
2015 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2016
2017 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
2018 arm_restrict_it.
2019
2020 2014-02-11 Renlin Li <Renlin.Li@arm.com>
2021
2022 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
2023 add_options_for_arm_vfp3.
2024
2025 2014-02-11 Jeff Law <law@redhat.com>
2026
2027 PR middle-end/54041
2028 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
2029 object with an undesirable mode.
2030
2031 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2032
2033 PR libgomp/60107
2034 * config/i386/sol2-9.h: New file.
2035 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
2036 *-*-solaris2.9*): Use it.
2037
2038 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2039
2040 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
2041 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
2042
2043 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2044
2045 * config/microblaze/microblaze.c: Extend mcpu version format
2046
2047 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
2048
2049 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
2050
2051 2014-02-10 Richard Henderson <rth@redhat.com>
2052
2053 PR target/59927
2054 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
2055 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
2056 ms-abi vs -mno-accumulate-outgoing-args.
2057 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
2058 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
2059 respect to ms-abi.
2060
2061 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
2062
2063 PR middle-end/60080
2064 * cfgexpand.c (expand_asm_operands): Attach source location to
2065 ASM_INPUT rtx objects.
2066 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
2067
2068 2014-02-10 Nick Clifton <nickc@redhat.com>
2069
2070 * config/mn10300/mn10300.c (popcount): New function.
2071 (mn10300_expand_prologue): Include saved registers in stack usage
2072 count.
2073
2074 2014-02-10 Jeff Law <law@redhat.com>
2075
2076 PR middle-end/52306
2077 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
2078 when changing the SET_DEST of a prior insn to avoid an input reload.
2079
2080 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
2081
2082 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
2083 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
2084 -mcall-openbsd, or -mcall-linux.
2085 (CC1_ENDIAN_BIG_SPEC): Remove.
2086 (CC1_ENDIAN_LITTLE_SPEC): Remove.
2087 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2088 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
2089 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
2090 and %cc1_endian_default.
2091 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2092
2093 2014-02-10 Richard Biener <rguenther@suse.de>
2094
2095 PR tree-optimization/60115
2096 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
2097 MEM_REF handling. Properly verify that the accesses are not
2098 out of the objects bound.
2099
2100 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2101
2102 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
2103 coretex to cortex.
2104
2105 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
2106
2107 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
2108 proper constants and fix formatting.
2109 (possible_polymorphic_call_targets): Fix formatting.
2110
2111 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
2112 Ilya Tocar <ilya.tocar@intel.com>
2113
2114 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
2115 (_mm512_loadu_epi32): Renamed into...
2116 (_mm512_loadu_si512): This.
2117 (_mm512_storeu_epi32): Renamed into...
2118 (_mm512_storeu_si512): This.
2119 (_mm512_maskz_ceil_ps): Removed.
2120 (_mm512_maskz_ceil_pd): Ditto.
2121 (_mm512_maskz_floor_ps): Ditto.
2122 (_mm512_maskz_floor_pd): Ditto.
2123 (_mm512_floor_round_ps): Ditto.
2124 (_mm512_floor_round_pd): Ditto.
2125 (_mm512_ceil_round_ps): Ditto.
2126 (_mm512_ceil_round_pd): Ditto.
2127 (_mm512_mask_floor_round_ps): Ditto.
2128 (_mm512_mask_floor_round_pd): Ditto.
2129 (_mm512_mask_ceil_round_ps): Ditto.
2130 (_mm512_mask_ceil_round_pd): Ditto.
2131 (_mm512_maskz_floor_round_ps): Ditto.
2132 (_mm512_maskz_floor_round_pd): Ditto.
2133 (_mm512_maskz_ceil_round_ps): Ditto.
2134 (_mm512_maskz_ceil_round_pd): Ditto.
2135 (_mm512_expand_pd): Ditto.
2136 (_mm512_expand_ps): Ditto.
2137 * config/i386/i386.c (ix86_builtins): Remove
2138 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
2139 (bdesc_args): Ditto.
2140 * config/i386/predicates.md (const1256_operand): New.
2141 (const_1_to_2_operand): Ditto.
2142 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2143 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2144 (*avx512pf_gatherpf<mode>sf): Ditto.
2145 (avx512pf_gatherpf<mode>df): Ditto.
2146 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2147 (*avx512pf_gatherpf<mode>df): Ditto.
2148 (avx512pf_scatterpf<mode>sf): Ditto.
2149 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2150 (*avx512pf_scatterpf<mode>sf): Ditto.
2151 (avx512pf_scatterpf<mode>df): Ditto.
2152 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2153 (*avx512pf_scatterpf<mode>df): Ditto.
2154 (avx512f_expand<mode>): Removed.
2155 (<shift_insn><mode>3<mask_name>): Change predicate type.
2156
2157 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2158
2159 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
2160 not at the end of datarefs vector use ordered_remove to avoid
2161 reordering datarefs vector.
2162
2163 PR c/59984
2164 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
2165 mark local addressable non-static vars as GOVD_PRIVATE
2166 instead of GOVD_LOCAL.
2167 * omp-low.c (lower_omp_for): Move gimple_bind_vars
2168 and BLOCK_VARS of gimple_bind_block to new_stmt rather
2169 than copying them.
2170
2171 PR middle-end/60092
2172 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
2173 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
2174 assume_aligned or alloc_align attributes.
2175 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
2176 arguments. Handle also assume_aligned and alloc_align attributes.
2177 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
2178 calls to functions with assume_aligned or alloc_align attributes.
2179 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
2180
2181 2014-02-08 Terry Guo <terry.guo@arm.com>
2182
2183 * doc/invoke.texi: Document ARM -march=armv7e-m.
2184
2185 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2186
2187 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
2188 flag on __cilkrts_rethrow builtin.
2189
2190 PR ipa/60026
2191 * ipa-cp.c (determine_versionability): Fail at -O0
2192 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
2193 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
2194
2195 Revert:
2196 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2197
2198 PR ipa/60026
2199 * tree-inline.c (copy_forbidden): Fail for
2200 __attribute__((optimize (0))) functions.
2201
2202 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2203
2204 * varpool.c: Include pointer-set.h.
2205 (varpool_remove_unreferenced_decls): Variables in other partitions
2206 will not be output; be however careful to not lose information
2207 about partitioning.
2208
2209 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2210
2211 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
2212 lookup in the vtable constructor.
2213
2214 2014-02-07 Jeff Law <law@redhat.com>
2215
2216 PR target/40977
2217 * config/m68k/m68k.md (ashldi_extsi): Turn into a
2218 define_insn_and_split.
2219
2220 * ipa-inline.c (inline_small_functions): Fix typos.
2221
2222 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2223
2224 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
2225 (s390_can_use_return_insn): Declare.
2226 * config/s390/s390.h (EPILOGUE_USES): Define.
2227 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
2228 instructions.
2229 (s390_chunkify_start): Handle return JUMP_LABELs.
2230 (s390_early_mach): Emit a main_pool instruction on the entry edge.
2231 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
2232 (s390_can_use_return_insn): New functions.
2233 (s390_fix_long_loop_prediction): Handle conditional returns.
2234 (TARGET_SET_UP_BY_PROLOGUE): Define.
2235 * config/s390/s390.md (ANY_RETURN): New code iterator.
2236 (*creturn, *csimple_return, return, simple_return): New patterns.
2237
2238 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2239
2240 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
2241 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
2242 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
2243 REG_CFA_RESTORE list when deciding not to restore a register.
2244
2245 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2246
2247 * config/s390/s390.c: Include tree-pass.h and context.h.
2248 (s390_early_mach): New function, split out from...
2249 (s390_emit_prologue): ...here.
2250 (pass_data_s390_early_mach): New pass structure.
2251 (pass_s390_early_mach): New class.
2252 (s390_option_override): Create and register early_mach pass.
2253 Move to end of file.
2254
2255 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2256
2257 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
2258 to match for the exit block.
2259
2260 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2261
2262 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
2263 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
2264 Reject misaligned operands.
2265
2266 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2267
2268 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
2269
2270 2014-02-07 Richard Biener <rguenther@suse.de>
2271
2272 PR middle-end/60092
2273 * gimple-low.c (lower_builtin_posix_memalign): New function.
2274 (lower_stmt): Call it to lower posix_memalign in a way
2275 to make alignment info accessible.
2276
2277 2014-02-07 Jakub Jelinek <jakub@redhat.com>
2278
2279 PR c++/60082
2280 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
2281 __builtin_setjmp_receiver.
2282
2283 2014-02-07 Richard Biener <rguenther@suse.de>
2284
2285 PR middle-end/60092
2286 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
2287 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
2288 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
2289 Handle BUILT_IN_POSIX_MEMALIGN.
2290 (find_func_clobbers): Likewise.
2291 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2292 (call_may_clobber_ref_p_1): Likewise.
2293
2294 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2295
2296 PR ipa/59918
2297 * ipa-devirt.c (record_target_from_binfo): Remove overactive
2298 sanity check.
2299
2300 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2301
2302 PR ipa/59469
2303 * lto-cgraph.c (lto_output_node): Use
2304 symtab_get_symbol_partitioning_class.
2305 (lto_output_varpool_node): likewise.
2306 (symtab_get_symbol_partitioning_class): Move here from
2307 lto/lto-partition.c
2308 * cgraph.h (symbol_partitioning_class): Likewise.
2309 (symtab_get_symbol_partitioning_class): Declare.
2310
2311 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2312
2313 * ggc.h (ggc_internal_cleared_alloc): New macro.
2314 * vec.h (vec_safe_copy): Handle memory stats.
2315 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
2316 * target-globals.c (save_target_globals): Likewise.
2317
2318 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2319
2320 PR target/60077
2321 * expr.c (emit_move_resolve_push): Export; be bit more selective
2322 on when to clear alias set.
2323 * expr.h (emit_move_resolve_push): Declare.
2324 * function.h (struct function): Add tail_call_marked.
2325 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
2326 * config/i386/i386-protos.h (ix86_expand_push): Remove.
2327 * config/i386/i386.md (TImode move expander): De not call
2328 ix86_expand_push.
2329 (FP push expanders): Preserve memory attributes.
2330 * config/i386/sse.md (push<mode>1): Remove.
2331 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
2332 (ix86_expand_push): Remove.
2333 * config/i386/mmx.md (push<mode>1): Remove.
2334
2335 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2336
2337 PR rtl-optimization/60030
2338 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
2339 lopart with paradoxical subreg before shifting it up by hprec.
2340
2341 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2342
2343 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
2344 Remove extra newline at end of file.
2345 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
2346 (arm_issue_rate): Handle cortexa57.
2347 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
2348 (cortex-a57.cortex-a53): Likewise.
2349
2350 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2351
2352 PR target/59575
2353 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
2354 don't record in REG_FRAME_RELATED_EXPR registers not set in that
2355 bitmask.
2356 (arm_expand_prologue): Adjust all callers.
2357 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
2358 info, registers also at the lowest numbered registers side. Use
2359 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
2360 XEXP.
2361
2362 PR debug/59992
2363 * var-tracking.c (adjust_mems): Before adding a SET to
2364 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
2365
2366 2014-02-06 Alan Modra <amodra@gmail.com>
2367
2368 PR target/60032
2369 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
2370 change SDmode to DDmode when lra_in_progress.
2371
2372 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2373
2374 PR middle-end/59150
2375 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
2376 free_data_ref on the dr first, and before goto again also set dr
2377 to the next dr. For simd_lane_access, free old datarefs[i] before
2378 overwriting it. For get_vectype_for_scalar_type failure, don't
2379 free_data_ref if simd_lane_access.
2380
2381 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
2382
2383 PR target/60062
2384 * tree.h (opts_for_fn): New inline function.
2385 (opt_for_fn): Define.
2386 * config/i386/i386.c (ix86_function_regparm): Use
2387 opt_for_fn (decl, optimize) instead of optimize.
2388
2389 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
2390
2391 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
2392 for SYMBOL_REF in large memory model.
2393
2394 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2395
2396 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
2397 and crypto support.
2398 (cortex-a57): Likewise.
2399 (cortex-a57.cortex-a53): Likewise.
2400
2401 2014-02-06 Yury Gribov <y.gribov@samsung.com>
2402 Kugan Vivekanandarajah <kuganv@linaro.org>
2403
2404 * config/arm/arm.c (arm_vector_alignment_reachable): Check
2405 unaligned_access.
2406 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
2407
2408 2014-02-06 Richard Biener <rguenther@suse.de>
2409
2410 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
2411 set_loop_copy and initialize_original_copy_tables.
2412
2413 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
2414
2415 * config/aarch64/aarch64-simd.md
2416 (aarch64_ashr_simddi): Change QI to SI.
2417
2418 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2419 Jakub Jelinek <jakub@redhat.com>
2420
2421 PR middle-end/60013
2422 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
2423 of the dataflow.
2424
2425 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2426
2427 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
2428 CODE_FOR_altivec_vpku[hw]um to
2429 CODE_FOR_altivec_vpku[hw]um_direct.
2430 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
2431 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
2432 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
2433 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
2434
2435 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2436
2437 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
2438 generation for -maltivec=be.
2439 (altivec_vsumsws): Simplify redundant test.
2440
2441 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2442
2443 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
2444 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
2445 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
2446 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
2447 gen_altivec_vpkuwum.
2448 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
2449 BYTES_BIG_ENDIAN.
2450 (altivec_vpks<VI_char>ss): Likewise.
2451 (altivec_vpks<VI_char>us): Likewise.
2452 (altivec_vpku<VI_char>us): Likewise.
2453 (altivec_vpku<VI_char>um): Likewise.
2454 (altivec_vpku<VI_char>um_direct): New (copy of
2455 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
2456 internal use).
2457 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
2458 target is little endian and -maltivec=be is not specified.
2459 (*altivec_vupkhs<VU_char>_direct): New (copy of
2460 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
2461 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
2462 target is little endian and -maltivec=be is not specified.
2463 (*altivec_vupkls<VU_char>_direct): New (copy of
2464 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
2465 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
2466 little endian and -maltivec=be is not specified.
2467 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
2468 little endian and -maltivec=be is not specified.
2469
2470 2014-02-05 Richard Henderson <rth@redhat.com>
2471
2472 PR debug/52727
2473 * combine-stack-adj.c: Revert r206943.
2474 * sched-int.h (struct deps_desc): Add last_args_size.
2475 * sched-deps.c (init_deps): Initialize it.
2476 (sched_analyze_insn): Add OUTPUT dependencies between insns that
2477 contain REG_ARGS_SIZE notes.
2478
2479 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2480
2481 * lto-cgraph.c (asm_nodes_output): Make global.
2482 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
2483 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
2484 (driver_handle_option): Handle OPT_fwpa.
2485
2486 2014-02-05 Jakub Jelinek <jakub@redhat.com>
2487
2488 PR ipa/59947
2489 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
2490 a comment typo and formatting issue. If odr_hash hasn't been
2491 created, return vNULL and set *completep to false.
2492
2493 PR middle-end/57499
2494 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
2495 bb with no successors.
2496
2497 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
2498
2499 PR target/59718
2500 * doc/invoke.texi (-march): Clarify documentation for ARM.
2501 (-mtune): Likewise.
2502 (-mcpu): Likewise.
2503
2504 2014-02-05 Richard Biener <rguenther@suse.de>
2505
2506 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
2507 when not vectorizing because of too many alias checks.
2508 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
2509 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
2510
2511 2014-02-05 Nick Clifton <nickc@redhat.com>
2512
2513 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
2514 accept extended registers in any mode when compiling for the MN10300.
2515
2516 2014-02-05 Yury Gribov <y.gribov@samsung.com>
2517
2518 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
2519 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
2520 sanitization attributes.
2521 (can_inline_edge_p): Likewise.
2522 (sanitize_attrs_match_for_inline_p): New function.
2523
2524 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2525
2526 * ipa-prop.c (detect_type_change): Shor circuit testing of
2527 type changes on THIS pointer.
2528
2529 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
2530
2531 PR target/59777
2532 * config/pa/pa.c (legitimize_tls_address): Return original address
2533 if not passed a SYMBOL_REF rtx.
2534 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
2535 addresses.
2536 (pa_emit_move_sequence): Simplify TLS source operands.
2537 (pa_legitimate_constant_p): Reject all TLS constants.
2538 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
2539 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
2540
2541 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2542
2543 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
2544 groups when we know they are controlled by LTO.
2545 * varasm.c (default_binds_local_p_1): If object is in other partition,
2546 it will be resolved locally.
2547
2548 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2549
2550 * config/host-linux.c (linux_gt_pch_use_address): Don't
2551 use SSIZE_MAX because it is not always defined.
2552
2553 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
2554
2555 PR bootstrap/59913
2556 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
2557 threshold for pseudo splitting.
2558 (update_ebb_live_info): Process call argument hard registers and
2559 hard registers from insn definition too.
2560 (max_small_class_regs_num): New constant.
2561 (inherit_in_ebb): Update live hard regs through EBBs. Update
2562 reloads_num only for small register classes. Don't split for
2563 outputs of jumps.
2564
2565 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
2566
2567 PR ipa/60058
2568 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
2569 is non-null.
2570
2571 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2572
2573 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
2574 visibility is safe.
2575
2576 2014-02-04 Marek Polacek <polacek@redhat.com>
2577
2578 * gdbinit.in (pel): Define.
2579
2580 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2581
2582 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
2583 behavior.
2584
2585 2014-02-04 Richard Biener <rguenther@suse.de>
2586
2587 PR lto/59723
2588 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
2589 in function context local.
2590 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
2591 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
2592 similar to LTO_imported_decl_ref.
2593
2594 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2595
2596 PR tree-optimization/60002
2597 * cgraphclones.c (build_function_decl_skip_args): Clear
2598 DECL_LANG_SPECIFIC.
2599
2600 PR tree-optimization/60023
2601 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
2602 false to gsi_replace.
2603 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
2604 has been in some EH region and vec_stmt could throw, add
2605 vec_stmt into the same EH region.
2606 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
2607 has no lhs, ignore it.
2608 * internal-fn.c (expand_MASK_LOAD): Likewise.
2609
2610 PR ipa/60026
2611 * tree-inline.c (copy_forbidden): Fail for
2612 __attribute__((optimize (0))) functions.
2613
2614 PR other/58712
2615 * omp-low.c (simd_clone_struct_copy): If from->inbranch
2616 is set, copy one less argument.
2617 (expand_simd_clones): Don't subtract clone_info->inbranch
2618 from simd_clone_struct_alloc argument.
2619
2620 PR rtl-optimization/57915
2621 * recog.c (simplify_while_replacing): If all unary/binary/relational
2622 operation arguments are constant, attempt to simplify those.
2623
2624 PR middle-end/59261
2625 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
2626 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
2627
2628 2014-02-04 Richard Biener <rguenther@suse.de>
2629
2630 PR tree-optimization/60012
2631 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
2632 TBAA disambiguation to all DDRs.
2633
2634 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2635
2636 PR target/59788
2637 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
2638 (LINK_SPEC): Use it for -shared, -shared-libgcc.
2639
2640 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2641
2642 PR ipa/59882
2643 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
2644
2645 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2646
2647 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
2648 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
2649
2650 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2651
2652 PR ipa/59831
2653 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
2654 to figure out targets of polymorphic calls with known decl.
2655 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2656 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
2657 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
2658 (get_polymorphic_call_info): ... here.
2659 (get_polymorphic_call_info_from_invariant): New function.
2660
2661 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2662
2663 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
2664 lookup via vtable pointer; check for type consistency
2665 and turn inconsitent facts into UNREACHABLE.
2666 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2667 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
2668 type inconsistent querries; return UNREACHABLE instead.
2669
2670 2014-02-03 Richard Henderson <rth@twiddle.net>
2671
2672 PR tree-opt/59924
2673 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
2674 already processed this node.
2675 (normalize_one_pred_1): Pass along mark_set.
2676 (normalize_one_pred): Create and destroy a pointer_set_t.
2677 (normalize_one_pred_chain): Likewise.
2678
2679 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
2680
2681 PR gcov-profile/58602
2682 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
2683
2684 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2685
2686 PR ipa/59831
2687 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
2688 -fno-devirtualize; try to devirtualize by the knowledge of
2689 virtual table pointer given by aggregate propagation.
2690 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2691 (ipa_print_node_jump_functions): Dump also offset that
2692 is relevant for polymorphic calls.
2693 (determine_known_aggregate_parts): Add arg_type parameter; use it
2694 instead of determining the type from pointer type.
2695 (ipa_compute_jump_functions_for_edge): Update call of
2696 determine_known_aggregate_parts.
2697 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
2698 (gimple_get_virt_method_for_binfo): ... here; simplify using
2699 vtable_pointer_value_to_vtable.
2700 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
2701 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
2702 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
2703 (vtable_pointer_value_to_vtable): Break out from ...; handle also
2704 POINTER_PLUS_EXPR.
2705 (vtable_pointer_value_to_binfo): ... here.
2706 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
2707
2708 2014-02-03 Teresa Johnson <tejohnson@google.com>
2709
2710 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
2711 redef of outer loop index variable.
2712
2713 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
2714
2715 PR c++/53017
2716 PR c++/59211
2717 * doc/extend.texi (Function Attributes): Typo.
2718
2719 2014-02-03 Cong Hou <congh@google.com>
2720
2721 PR tree-optimization/60000
2722 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
2723 if the vectorized statement is a store. A store statement can only
2724 appear at the end of pattern statements.
2725
2726 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2727
2728 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
2729 (ix86_option_override_internal): Default long double to 64-bit for
2730 32-bit Bionic and to 128-bit for 64-bit Bionic.
2731
2732 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
2733 TARGET_LONG_DOUBLE_128 is true.
2734 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
2735
2736 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
2737 (mlong-double-64): Negate -mlong-double-128.
2738 (mlong-double-128): New option.
2739
2740 * config/i386/i386-c.c (ix86_target_macros): Define
2741 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
2742
2743 * doc/invoke.texi: Document -mlong-double-128.
2744
2745 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2746
2747 PR rtl-optimization/60024
2748 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
2749
2750 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
2751
2752 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
2753
2754 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2755
2756 PR rtl-optimization/57662
2757 * sel-sched.c (code_motion_path_driver): Do not mark already not
2758 existing blocks in the visiting bitmap.
2759
2760 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2761
2762 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
2763 on the insn being emitted.
2764
2765 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
2766 Will Deacon <will.deacon@arm.com>
2767
2768 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
2769
2770 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2771
2772 * config/arm/arm-tables.opt: Regenerate.
2773
2774 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2775
2776 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
2777 for vector types other than V16QImode.
2778 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
2779 define_expand, and call altivec_expand_vec_perm_le when producing
2780 code with little endian element order.
2781 (*altivec_vperm_<mode>_internal): New insn having previous
2782 behavior of altivec_vperm_<mode>.
2783 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
2784 altivec_expand_vec_perm_le when producing code with little endian
2785 element order.
2786 (*altivec_vperm_<mode>_uns_internal): New insn having previous
2787 behavior of altivec_vperm_<mode>_uns.
2788
2789 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2790
2791 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
2792 (altivec_vsumsws): Add handling for -maltivec=be with a little
2793 endian target.
2794 (altivec_vsumsws_direct): New.
2795 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
2796 gen_altivec_vsumsws.
2797
2798 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2799
2800 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
2801 vtable_pointer_value_to_binfo): New functions.
2802 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
2803 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
2804
2805 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
2806
2807 * config/nios2/nios2.md (load_got_register): Initialize GOT
2808 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
2809 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
2810
2811 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2812
2813 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
2814 preserverd by passthrough, do not propagate the type.
2815
2816 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2817
2818 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
2819 (mips_atomic_assign_expand_fenv): New function.
2820 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
2821
2822 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2823
2824 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
2825 (__builtin_mips_set_fcsr): Likewise.
2826 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
2827 MIPS_USI_FTYPE_VOID.
2828 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
2829 (mips16_expand_set_fcsr): Likewise.
2830 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
2831 (mips16_set_fcsr_stub): Likewise.
2832 (mips16_get_fcsr_one_only_stub): New class.
2833 (mips16_set_fcsr_one_only_stub): Likewise.
2834 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
2835 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
2836 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
2837 (hard_float): New availability predicate.
2838 (mips_builtins): Add get_fcsr and set_fcsr.
2839 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
2840 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
2841 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
2842 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
2843 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
2844 patterns.
2845
2846 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2847
2848 * config/mips/mips.c (mips_one_only_stub): New class.
2849 (mips_need_mips16_rdhwr_p): Replace with...
2850 (mips16_rdhwr_stub): ...this new variable.
2851 (mips16_stub_call_address): New function.
2852 (mips16_rdhwr_one_only_stub): New class.
2853 (mips_expand_thread_pointer): Use mips16_stub_call_address.
2854 (mips_output_mips16_rdhwr): Delete.
2855 (mips_finish_stub): New function.
2856 (mips_code_end): Use it to handle rdhwr stubs.
2857
2858 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
2859
2860 PR target/60017
2861 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
2862 when calculating size of integer atomic types.
2863
2864 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
2865
2866 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
2867
2868 2014-02-01 Jakub Jelinek <jakub@redhat.com>
2869
2870 PR tree-optimization/60003
2871 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
2872 * profile.c (branch_prob): Use gimple_call_builtin_p
2873 to check for BUILT_IN_SETJMP_RECEIVER.
2874 * tree-inline.c (copy_bb): Call notice_special_calls.
2875
2876 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
2877
2878 PR bootstrap/59985
2879 * lra-constraints.c (process_alt_operands): Update reload_sum only
2880 on the first pass.
2881
2882 2014-01-31 Richard Henderson <rth@redhat.com>
2883
2884 PR middle-end/60004
2885 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
2886 until after else_eh is processed.
2887
2888 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2889
2890 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
2891 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
2892 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
2893 in smmintrin.h, remove them.
2894 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
2895 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
2896 * config/i386/i386.md (ROUND_SAE): Fix value.
2897 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
2898 (const48_operand): New.
2899 * config/i386/subst.md (round), (round_expand): Use
2900 const_4_or_8_to_11_operand.
2901 (round_saeonly), (round_saeonly_expand): Use const48_operand.
2902
2903 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2904
2905 * config/i386/constraints.md (Yk): Swap meaning with k.
2906 * config/i386/i386.md (movhi_internal): Change Yk to k.
2907 (movqi_internal): Ditto.
2908 (*k<logic><mode>): Ditto.
2909 (*andhi_1): Ditto.
2910 (*andqi_1): Ditto.
2911 (kandn<mode>): Ditto.
2912 (*<code>hi_1): Ditto.
2913 (*<code>qi_1): Ditto.
2914 (kxnor<mode>): Ditto.
2915 (kortestzhi): Ditto.
2916 (kortestchi): Ditto.
2917 (kunpckhi): Ditto.
2918 (*one_cmplhi2_1): Ditto.
2919 (*one_cmplqi2_1): Ditto.
2920 * config/i386/sse.md (): Change k to Yk.
2921 (avx512f_load<mode>_mask): Ditto.
2922 (avx512f_blendm<mode>): Ditto.
2923 (avx512f_store<mode>_mask): Ditto.
2924 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
2925 (avx512f_storedqu<mode>_mask): Ditto.
2926 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
2927 Ditto.
2928 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
2929 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
2930 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
2931 (avx512f_maskcmp<mode>3): Ditto.
2932 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
2933 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
2934 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
2935 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
2936 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
2937 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
2938 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
2939 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
2940 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
2941 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
2942 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
2943 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
2944 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
2945 (vec_extract_lo_<mode>_maskm): Ditto.
2946 (vec_extract_hi_<mode>_maskm): Ditto.
2947 (avx512f_vternlog<mode>_mask): Ditto.
2948 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
2949 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
2950 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
2951 (avx512f_<code>v8div16qi2_mask): Ditto.
2952 (avx512f_<code>v8div16qi2_mask_store): Ditto.
2953 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
2954 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
2955 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
2956 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
2957 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2958 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2959 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2960 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2961 (avx512cd_maskb_vec_dupv8di): Ditto.
2962 (avx512cd_maskw_vec_dupv16si): Ditto.
2963 (avx512f_vpermi2var<mode>3_maskz): Ditto.
2964 (avx512f_vpermi2var<mode>3_mask): Ditto.
2965 (avx512f_vpermi2var<mode>3_mask): Ditto.
2966 (avx512f_vpermt2var<mode>3_maskz): Ditto.
2967 (*avx512f_gathersi<mode>): Ditto.
2968 (*avx512f_gathersi<mode>_2): Ditto.
2969 (*avx512f_gatherdi<mode>): Ditto.
2970 (*avx512f_gatherdi<mode>_2): Ditto.
2971 (*avx512f_scattersi<mode>): Ditto.
2972 (*avx512f_scatterdi<mode>): Ditto.
2973 (avx512f_compress<mode>_mask): Ditto.
2974 (avx512f_compressstore<mode>_mask): Ditto.
2975 (avx512f_expand<mode>_mask): Ditto.
2976 * config/i386/subst.md (mask): Change k to Yk.
2977 (mask_scalar_merge): Ditto.
2978 (sd): Ditto.
2979
2980 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
2981
2982 * doc/extend.texi (Vector Extensions): Document ?: in C++.
2983
2984 2014-01-31 Richard Biener <rguenther@suse.de>
2985
2986 PR middle-end/59990
2987 * builtins.c (fold_builtin_memory_op): Make sure to not
2988 use a floating-point mode or a boolean or enumeral type for
2989 the copy operation.
2990
2991 2014-01-30 DJ Delorie <dj@redhat.com>
2992
2993 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
2994 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
2995 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
2996 whenever main() has an epilogue.
2997
2998 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2999
3000 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
3001 unused variable "field".
3002 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
3003 (vsx_mergeh_<mode>): Likewise.
3004 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
3005 (altivec_vmrghh): Likewise.
3006 (altivec_vmrghw): Likewise.
3007 (altivec_vmrglb): Likewise.
3008 (altivec_vmrglh): Likewise.
3009 (altivec_vmrglw): Likewise.
3010 (altivec_vspltb): Add missing uses.
3011 (altivec_vsplth): Likewise.
3012 (altivec_vspltw): Likewise.
3013 (altivec_vspltsf): Likewise.
3014
3015 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3016
3017 PR target/59923
3018 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
3019 frame related instructions.
3020
3021 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
3022
3023 PR rtl-optimization/59959
3024 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
3025 any reload of register whose subreg is invalid.
3026
3027 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3028
3029 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
3030 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
3031 Add missing return type - void.
3032
3033 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3034
3035 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
3036 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
3037 remove element index adjustment for endian (now handled in vsx.md
3038 and altivec.md).
3039 (altivec_expand_vec_perm_const): Use
3040 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
3041 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
3042 (vsx_xxspltw_<mode>): Adjust element index for little endian.
3043 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
3044 define_expand and a new define_insn *altivec_vspltb_internal;
3045 adjust for -maltivec=be on a little endian target.
3046 (altivec_vspltb_direct): New.
3047 (altivec_vsplth): Divide into a define_expand and a new
3048 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
3049 little endian target.
3050 (altivec_vsplth_direct): New.
3051 (altivec_vspltw): Divide into a define_expand and a new
3052 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
3053 little endian target.
3054 (altivec_vspltw_direct): New.
3055 (altivec_vspltsf): Divide into a define_expand and a new
3056 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
3057 a little endian target.
3058
3059 2014-01-30 Richard Biener <rguenther@suse.de>
3060
3061 PR tree-optimization/59993
3062 * tree-ssa-forwprop.c (associate_pointerplus): Check we
3063 can propagate form the earlier stmt and avoid the transform
3064 when the intermediate result is needed.
3065
3066 2014-01-30 Alangi Derick <alangiderick@gmail.com>
3067
3068 * README.Portability: Fix typo.
3069
3070 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
3071
3072 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
3073 comparison_operator with ordered_comparison_operator.
3074
3075 2014-01-30 Nick Clifton <nickc@redhat.com>
3076
3077 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
3078 Rename to mn10300_store_multiple_regs.
3079 * config/mn10300/mn10300.c: Likewise.
3080 * config/mn10300/mn10300.md (store_movm): Fix typo: call
3081 store_multiple_regs.
3082 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
3083 Call mn10300_store_multiple_regs.
3084
3085 2014-01-30 Nick Clifton <nickc@redhat.com>
3086 DJ Delorie <dj@redhat.com>
3087
3088 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
3089 %fp 2 to keep registers after it properly word-aligned.
3090 (rl78_alloc_physical_registers_umul): Handle the case where both
3091 input operands are the same.
3092
3093 2014-01-30 Richard Biener <rguenther@suse.de>
3094
3095 PR tree-optimization/59903
3096 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
3097 check properly.
3098
3099 2014-01-30 Jason Merrill <jason@redhat.com>
3100
3101 PR c++/59633
3102 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
3103
3104 PR c++/59645
3105 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
3106
3107 2014-01-30 Richard Biener <rguenther@suse.de>
3108
3109 PR tree-optimization/59951
3110 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
3111
3112 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
3113
3114 PR target/59784
3115 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
3116 SFmode to DFmode case.
3117
3118 2014-01-29 DJ Delorie <dj@redhat.com>
3119
3120 * config/msp430/msp430.opt (-minrt): New.
3121 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
3122 if -minrt given.
3123 (ENDFILE_SPEC): Likewise.
3124
3125 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
3126
3127 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
3128 (estimate_function_body_sizes): Use it.
3129
3130 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
3131
3132 PR c++/58561
3133 * dwarf2out.c (is_cxx_auto): New.
3134 (is_base_type): Use it.
3135 (gen_type_die_with_usage): Likewise.
3136
3137 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3138
3139 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
3140 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
3141 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
3142 -maltivec=be with LE targets.
3143 (vsx_mergeh_<mode>): Likewise.
3144 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
3145 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
3146 (altivec_vmrghb): Replace with define_expand and new
3147 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
3148 (altivec_vmrghb_direct): New define_insn.
3149 (altivec_vmrghh): Replace with define_expand and new
3150 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
3151 (altivec_vmrghh_direct): New define_insn.
3152 (altivec_vmrghw): Replace with define_expand and new
3153 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
3154 (altivec_vmrghw_direct): New define_insn.
3155 (*altivec_vmrghsf): Adjust for endianness.
3156 (altivec_vmrglb): Replace with define_expand and new
3157 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
3158 (altivec_vmrglb_direct): New define_insn.
3159 (altivec_vmrglh): Replace with define_expand and new
3160 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
3161 (altivec_vmrglh_direct): New define_insn.
3162 (altivec_vmrglw): Replace with define_expand and new
3163 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
3164 (altivec_vmrglw_direct): New define_insn.
3165 (*altivec_vmrglsf): Adjust for endianness.
3166 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3167 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3168 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3169 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3170 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3171 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3172 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3173 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3174
3175 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
3176
3177 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
3178 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
3179 whitespace.
3180
3181 2014-01-29 Richard Biener <rguenther@suse.de>
3182
3183 PR tree-optimization/58742
3184 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
3185 associate_pointerplus_align.
3186 (associate_pointerplus_diff): New function.
3187 (associate_pointerplus): Likewise. Call associate_pointerplus_align
3188 and associate_pointerplus_diff.
3189
3190 2014-01-29 Richard Biener <rguenther@suse.de>
3191
3192 * lto-streamer.h (LTO_major_version): Bump to 3.
3193 (LTO_minor_version): Reset to 0.
3194
3195 2014-01-29 Renlin Li <Renlin.Li@arm.com>
3196
3197 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
3198 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
3199 (arm_file_start): Generate correct asm header for armv7ve.
3200 * config/arm/bpabi.h: Add multilib support for armv7ve.
3201 * config/arm/driver-arm.c: Change the architectures of cortex-a7
3202 and cortex-a15 to armv7ve.
3203 * config/arm/t-aprofile: Add multilib support for armv7ve.
3204 * doc/invoke.texi: Document -march=armv7ve.
3205
3206 2014-01-29 Richard Biener <rguenther@suse.de>
3207
3208 PR tree-optimization/58742
3209 * tree-ssa-forwprop.c (associate_plusminus): Return true
3210 if we changed sth, defer EH cleanup to ...
3211 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
3212 (simplify_mult): New function.
3213
3214 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3215
3216 PR middle-end/59917
3217 PR tree-optimization/59920
3218 * tree.c (build_common_builtin_nodes): Remove
3219 __builtin_setjmp_dispatcher initialization.
3220 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
3221 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
3222 instead of gsi_after_labels + manually skipping debug stmts.
3223 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
3224 ignore bbs with IFN_ABNORMAL_DISPATCHER.
3225 * tree-inline.c (copy_edges_for_bb): Remove
3226 can_make_abnormal_goto argument, instead add abnormal_goto_dest
3227 argument. Ignore computed_goto_p stmts. Don't call
3228 make_abnormal_goto_edges. If a call might need abnormal edges
3229 for non-local gotos, see if it already has an edge to
3230 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
3231 with true argument, don't do anything then, otherwise add
3232 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
3233 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
3234 caller.
3235 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
3236 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
3237 (lower_stmt): Don't set data->calls_builtin_setjmp.
3238 (lower_builtin_setjmp): Adjust comment.
3239 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
3240 * tree-cfg.c (found_computed_goto): Remove.
3241 (factor_computed_gotos): Remove.
3242 (make_goto_expr_edges): Return bool, true for computed gotos.
3243 Don't call make_abnormal_goto_edges.
3244 (build_gimple_cfg): Don't set found_computed_goto, don't call
3245 factor_computed_gotos.
3246 (computed_goto_p): No longer static.
3247 (make_blocks): Don't set found_computed_goto.
3248 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
3249 (make_edges): If make_goto_expr_edges returns true, push bb
3250 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
3251 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
3252 vector. Record mapping between bbs and OpenMP regions if there
3253 are any, adjust make_gimple_omp_edges caller. Call
3254 handle_abnormal_edges.
3255 (make_abnormal_goto_edges): Remove.
3256 * tree-cfg.h (make_abnormal_goto_edges): Remove.
3257 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
3258 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
3259 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
3260 * internal-fn.def (ABNORMAL_DISPATCHER): New.
3261 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
3262 filling *region also set *region_idx to (*region)->entry->index.
3263
3264 PR other/58712
3265 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
3266 For REGs set ORIGINAL_REGNO.
3267
3268 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
3269
3270 * doc/md.texi: Mention that a target shouldn't implement
3271 vec_widen_(s|u)mul_even/odd pair if it is less efficient
3272 than hi/lo pair.
3273
3274 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3275
3276 PR tree-optimization/59594
3277 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
3278 a copy of the datarefs vector rather than the vector itself.
3279
3280 2014-01-28 Jason Merrill <jason@redhat.com>
3281
3282 PR c++/53756
3283 * dwarf2out.c (auto_die): New static.
3284 (gen_type_die_with_usage): Handle C++1y 'auto'.
3285 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
3286 on definition.
3287
3288 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
3289
3290 PR target/59672
3291 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
3292 (SPEC_X32): Likewise.
3293 (SPEC_64): Likewise.
3294 * config/i386/i386.c (ix86_option_override_internal): Turn off
3295 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
3296 for TARGET_16BIT.
3297 (x86_file_start): Output .code16gcc for TARGET_16BIT.
3298 * config/i386/i386.h (TARGET_16BIT): New macro.
3299 (TARGET_16BIT_P): Likewise.
3300 * config/i386/i386.opt: Add m16.
3301 * doc/invoke.texi: Document -m16.
3302
3303 2014-01-28 Jakub Jelinek <jakub@redhat.com>
3304
3305 PR preprocessor/59935
3306 * input.c (location_get_source_line): Bail out on when line number
3307 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
3308
3309 2014-01-28 Richard Biener <rguenther@suse.de>
3310
3311 PR tree-optimization/58742
3312 * tree-ssa-forwprop.c (associate_plusminus): Handle
3313 pointer subtraction of the form (T)(P + A) - (T)P.
3314
3315 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3316
3317 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
3318 at const_int_cost.
3319
3320 2014-01-28 Richard Biener <rguenther@suse.de>
3321
3322 Revert
3323 2014-01-28 Richard Biener <rguenther@suse.de>
3324
3325 PR rtl-optimization/45364
3326 PR rtl-optimization/59890
3327 * var-tracking.c (local_get_addr_clear_given_value): Handle
3328 already cleared slot.
3329 (val_reset): Handle not allocated local_get_addr_cache.
3330 (vt_find_locations): Use post-order on the inverted CFG.
3331
3332 2014-01-28 Richard Biener <rguenther@suse.de>
3333
3334 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
3335
3336 2014-01-28 Richard Biener <rguenther@suse.de>
3337
3338 PR rtl-optimization/45364
3339 PR rtl-optimization/59890
3340 * var-tracking.c (local_get_addr_clear_given_value): Handle
3341 already cleared slot.
3342 (val_reset): Handle not allocated local_get_addr_cache.
3343 (vt_find_locations): Use post-order on the inverted CFG.
3344
3345 2014-01-28 Alan Modra <amodra@gmail.com>
3346
3347 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
3348 * configure.ac <recursive call for build != host>: Define
3349 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
3350 and LD_FOR_BUILD too.
3351 * configure: Regenerate.
3352
3353 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
3354
3355 * config/i386/i386.c (get_builtin_code_for_version): Separate
3356 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
3357 Broadwell from Haswell.
3358
3359 2014-01-27 Steve Ellcey <sellcey@mips.com>
3360
3361 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
3362 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
3363 * config/mips/mips.c (mips_option_override): Change setting
3364 of TARGET_DSP.
3365 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
3366 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
3367 Change from Mask to Var.
3368
3369 2014-01-27 Jeff Law <law@redhat.com>
3370
3371 * ipa-inline.c (inline_small_functions): Fix typo.
3372
3373 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
3374
3375 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
3376 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
3377 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
3378 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
3379 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
3380 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
3381 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
3382 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
3383 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
3384 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
3385 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
3386 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
3387 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
3388 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
3389 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
3390 (_mm512_storeu_epi64): Ditto.
3391 (_mm512_cmpge_epi32_mask): Ditto.
3392 (_mm512_cmpge_epu32_mask): Ditto.
3393 (_mm512_cmpge_epi64_mask): Ditto.
3394 (_mm512_cmpge_epu64_mask): Ditto.
3395 (_mm512_cmple_epi32_mask): Ditto.
3396 (_mm512_cmple_epu32_mask): Ditto.
3397 (_mm512_cmple_epi64_mask): Ditto.
3398 (_mm512_cmple_epu64_mask): Ditto.
3399 (_mm512_cmplt_epi32_mask): Ditto.
3400 (_mm512_cmplt_epu32_mask): Ditto.
3401 (_mm512_cmplt_epi64_mask): Ditto.
3402 (_mm512_cmplt_epu64_mask): Ditto.
3403 (_mm512_cmpneq_epi32_mask): Ditto.
3404 (_mm512_cmpneq_epu32_mask): Ditto.
3405 (_mm512_cmpneq_epi64_mask): Ditto.
3406 (_mm512_cmpneq_epu64_mask): Ditto.
3407 (_mm512_expand_pd): Ditto.
3408 (_mm512_expand_ps): Ditto.
3409 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
3410 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
3411 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
3412 * config/i386/i386.c (ix86_builtins): Add
3413 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
3414 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
3415 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
3416 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
3417 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
3418 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
3419 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
3420 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
3421 IX86_BUILTIN_PMOVUSQW512_MEM.
3422 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
3423 __builtin_ia32_pmovsqd512mem_mask,
3424 __builtin_ia32_pmovqd512mem_mask,
3425 __builtin_ia32_pmovusqw512mem_mask,
3426 __builtin_ia32_pmovsqw512mem_mask,
3427 __builtin_ia32_pmovqw512mem_mask,
3428 __builtin_ia32_pmovusdw512mem_mask,
3429 __builtin_ia32_pmovsdw512mem_mask,
3430 __builtin_ia32_pmovdw512mem_mask,
3431 __builtin_ia32_pmovqb512mem_mask,
3432 __builtin_ia32_pmovusqb512mem_mask,
3433 __builtin_ia32_pmovsqb512mem_mask,
3434 __builtin_ia32_pmovusdb512mem_mask,
3435 __builtin_ia32_pmovsdb512mem_mask,
3436 __builtin_ia32_pmovdb512mem_mask.
3437 (bdesc_args): Add __builtin_ia32_expanddf512,
3438 __builtin_ia32_expandsf512.
3439 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
3440 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
3441 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
3442 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
3443 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
3444 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
3445 (avx512f_<code>v8div16qi2_mask_store): This.
3446 (avx512f_expand<mode>): New.
3447
3448 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
3449
3450 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
3451 New.
3452 (_mm512_mask_prefetch_i64gather_pd): Ditto.
3453 (_mm512_prefetch_i32scatter_pd): Ditto.
3454 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
3455 (_mm512_prefetch_i64scatter_pd): Ditto.
3456 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
3457 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
3458 (_mm512_mask_prefetch_i64gather_ps): Ditto.
3459 (_mm512_prefetch_i32scatter_ps): Ditto.
3460 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
3461 (_mm512_prefetch_i64scatter_ps): Ditto.
3462 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
3463 * config/i386/i386-builtin-types.def: Define
3464 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
3465 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
3466 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
3467 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
3468 IX86_BUILTIN_SCATTERPFQPD.
3469 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
3470 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
3471 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
3472 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
3473 __builtin_ia32_scatterpfqps.
3474 (ix86_expand_builtin): Expand new built-ins.
3475 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
3476 fix memory access data type.
3477 (*avx512pf_gatherpf<mode>_mask): Ditto.
3478 (*avx512pf_gatherpf<mode>): Ditto.
3479 (avx512pf_scatterpf<mode>): Ditto.
3480 (*avx512pf_scatterpf<mode>_mask): Ditto.
3481 (*avx512pf_scatterpf<mode>): Ditto.
3482 (GATHER_SCATTER_SF_MEM_MODE): New.
3483 (avx512pf_gatherpf<mode>df): Ditto.
3484 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3485 (*avx512pf_scatterpf<mode>df): Ditto.
3486
3487 2014-01-27 Jakub Jelinek <jakub@redhat.com>
3488
3489 PR bootstrap/59934
3490 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
3491 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
3492 reached.
3493
3494 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3495
3496 * common/config/arm/arm-common.c
3497 (arm_rewrite_mcpu): Handle multiple names.
3498 * config/arm/arm.h
3499 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3500
3501 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3502
3503 * gimple-builder.h (create_gimple_tmp): Delete.
3504
3505 2014-01-27 Christian Bruel <christian.bruel@st.com>
3506
3507 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
3508 words comparisons.
3509
3510 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
3511
3512 * config/pa/pa.md (call): Generate indirect long calls to non-local
3513 functions when outputing 32-bit code.
3514 (call_value): Likewise except for special call to buggy powf function.
3515
3516 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
3517 portable runtime and PIC indirect calls.
3518 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
3519 and PIC call sequences. Use ldo instead of blr to set return register
3520 in PIC call sequence.
3521
3522 2014-01-25 Walter Lee <walt@tilera.com>
3523
3524 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
3525 avoid clobbering a live register.
3526
3527 2014-01-25 Walter Lee <walt@tilera.com>
3528
3529 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
3530 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
3531 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
3532 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
3533
3534 2014-01-25 Walter Lee <walt@tilera.com>
3535
3536 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
3537 arguments on even registers.
3538 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
3539 STACK_BOUNDARY.
3540 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
3541 (BIGGEST_ALIGNMENT): Ditto.
3542 (BIGGEST_FIELD_ALIGNMENT): Ditto.
3543
3544 2014-01-25 Walter Lee <walt@tilera.com>
3545
3546 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
3547 insns before bundling.
3548 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
3549
3550 2014-01-25 Walter Lee <walt@tilera.com>
3551
3552 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
3553 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
3554 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
3555
3556 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3557
3558 * config/mips/constraints.md (kl): Delete.
3559 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
3560 define expands, using...
3561 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
3562 instructions for MIPS16.
3563 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
3564 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
3565
3566 2014-01-25 Walter Lee <walt@tilera.com>
3567
3568 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
3569 (clzdi2): Ditto.
3570 (ffsdi2): Ditto.
3571
3572 2014-01-25 Walter Lee <walt@tilera.com>
3573
3574 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
3575 (TARGET_EXPAND_TO_RTL_HOOK): Define.
3576
3577 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3578
3579 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
3580 Handle XOR.
3581
3582 2014-01-25 Jakub Jelinek <jakub@redhat.com>
3583
3584 * print-rtl.c (in_call_function_usage): New var.
3585 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
3586 EXPR_LIST mode as mode and not as reg note name.
3587
3588 PR middle-end/59561
3589 * cfgloopmanip.c (copy_loop_info): If
3590 loop->warned_aggressive_loop_optimizations, make sure
3591 the flag is set in target loop too.
3592
3593 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
3594
3595 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
3596 flag_cilkplus.
3597 * builtins.def: Likewise.
3598 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
3599 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
3600 * ira.c (ira_setup_eliminable_regset): Likewise.
3601 * omp-low.c (gate_expand_omp): Likewise.
3602 (execute_lower_omp): Likewise.
3603 (diagnose_sb_0): Likewise.
3604 (gate_diagnose_omp_blocks): Likewise.
3605 (simd_clone_clauses_extract): Likewise.
3606 (gate): Likewise.
3607
3608 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3609
3610 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
3611 correction for little endian...
3612 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
3613 here.
3614
3615 2014-01-24 Jeff Law <law@redhat.com>
3616
3617 PR tree-optimization/59919
3618 * tree-vrp.c (find_assert_locations_1): Do not register asserts
3619 for non-returning calls.
3620
3621 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
3622
3623 * common/config/aarch64/aarch64-common.c
3624 (aarch64_rewrite_mcpu): Handle multiple names.
3625 * config/aarch64/aarch64.h
3626 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3627
3628 2014-01-24 Dodji Seketeli <dodji@redhat.com>
3629
3630 * input.c (add_file_to_cache_tab): Handle the case where fopen
3631 returns NULL.
3632
3633 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
3634
3635 PR target/59929
3636 * config/i386/i386.md (pushsf splitter): Get stack adjustment
3637 from push operand if code of push isn't PRE_DEC.
3638
3639 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3640
3641 PR target/59909
3642 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
3643 -mquad-memory-atomic. Update -mquad-memory documentation to say
3644 it is only used for non-atomic loads/stores.
3645
3646 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
3647 -mquad-memory or -mquad-memory-atomic switches.
3648
3649 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
3650 -mquad-memory-atomic to ISA 2.07 support.
3651
3652 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
3653 to separate support of normal quad word memory operations (ldq, stq)
3654 from the atomic quad word memory operations.
3655
3656 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
3657 support to separate non-atomic quad word operations from atomic
3658 quad word operations. Disable non-atomic quad word operations in
3659 little endian mode so that we don't have to swap words after the
3660 load and before the store.
3661 (quad_load_store_p): Add comment about atomic quad word support.
3662 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
3663 options printed with -mdebug=reg.
3664
3665 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
3666 -mquad-memory-atomic as the test for whether we have quad word
3667 atomic instructions.
3668 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
3669 or -mp8-vector are used, allow byte/half-word atomic operations.
3670
3671 * config/rs6000/sync.md (load_lockedti): Insure that the address
3672 is a proper indexed or indirect address for the lqarx instruction.
3673 On little endian systems, swap the hi/lo registers after the lqarx
3674 instruction.
3675 (load_lockedpti): Use indexed_or_indirect_operand predicate to
3676 insure the address is valid for the lqarx instruction.
3677 (store_conditionalti): Insure that the address is a proper indexed
3678 or indirect address for the stqcrx. instruction. On little endian
3679 systems, swap the hi/lo registers before doing the stqcrx.
3680 instruction.
3681 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
3682 insure the address is valid for the stqcrx. instruction.
3683
3684 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
3685 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
3686 type of quad memory support is available.
3687
3688 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
3689
3690 PR regression/59915
3691 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
3692 there is a danger of looping.
3693
3694 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
3695
3696 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3697 force flag_ira_loop_pressure if set via command line.
3698
3699 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3700
3701 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
3702 (ashr_simd): New builtin handling DI mode.
3703 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
3704 (aarch64_sshr_simddi): New match pattern.
3705 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
3706 (vshrd_n_s64): Likewise.
3707 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
3708
3709 2014-01-23 Nick Clifton <nickc@redhat.com>
3710
3711 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
3712 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
3713 favour of mcu specific scripts.
3714 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
3715 430x multilibs.
3716
3717 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3718 Alex Velenko <Alex.Velenko@arm.com>
3719
3720 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
3721 (vaddv_s16): Likewise.
3722 (vaddv_s32): Likewise.
3723 (vaddv_u8): Likewise.
3724 (vaddv_u16): Likewise.
3725 (vaddv_u32): Likewise.
3726 (vaddvq_s8): Likewise.
3727 (vaddvq_s16): Likewise.
3728 (vaddvq_s32): Likewise.
3729 (vaddvq_s64): Likewise.
3730 (vaddvq_u8): Likewise.
3731 (vaddvq_u16): Likewise.
3732 (vaddvq_u32): Likewise.
3733 (vaddvq_u64): Likewise.
3734 (vaddv_f32): Likewise.
3735 (vaddvq_f32): Likewise.
3736 (vaddvq_f64): Likewise.
3737 (vmaxv_f32): Likewise.
3738 (vmaxv_s8): Likewise.
3739 (vmaxv_s16): Likewise.
3740 (vmaxv_s32): Likewise.
3741 (vmaxv_u8): Likewise.
3742 (vmaxv_u16): Likewise.
3743 (vmaxv_u32): Likewise.
3744 (vmaxvq_f32): Likewise.
3745 (vmaxvq_f64): Likewise.
3746 (vmaxvq_s8): Likewise.
3747 (vmaxvq_s16): Likewise.
3748 (vmaxvq_s32): Likewise.
3749 (vmaxvq_u8): Likewise.
3750 (vmaxvq_u16): Likewise.
3751 (vmaxvq_u32): Likewise.
3752 (vmaxnmv_f32): Likewise.
3753 (vmaxnmvq_f32): Likewise.
3754 (vmaxnmvq_f64): Likewise.
3755 (vminv_f32): Likewise.
3756 (vminv_s8): Likewise.
3757 (vminv_s16): Likewise.
3758 (vminv_s32): Likewise.
3759 (vminv_u8): Likewise.
3760 (vminv_u16): Likewise.
3761 (vminv_u32): Likewise.
3762 (vminvq_f32): Likewise.
3763 (vminvq_f64): Likewise.
3764 (vminvq_s8): Likewise.
3765 (vminvq_s16): Likewise.
3766 (vminvq_s32): Likewise.
3767 (vminvq_u8): Likewise.
3768 (vminvq_u16): Likewise.
3769 (vminvq_u32): Likewise.
3770 (vminnmv_f32): Likewise.
3771 (vminnmvq_f32): Likewise.
3772 (vminnmvq_f64): Likewise.
3773
3774 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3775
3776 * config/aarch64/aarch64-simd.md
3777 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
3778 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
3779 (*aarch64_mul3_elt<mode>): Likewise.
3780 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
3781 (*aarch64_mul3_elt_to_64v2df): Likewise.
3782 (*aarch64_mla_elt<mode>): Likewise.
3783 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
3784 (*aarch64_mls_elt<mode>): Likewise.
3785 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
3786 (*aarch64_fma4_elt<mode>): Likewise.
3787 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
3788 (*aarch64_fma4_elt_to_64v2df): Likewise.
3789 (*aarch64_fnma4_elt<mode>): Likewise.
3790 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3791 (*aarch64_fnma4_elt_to_64v2df): Likewise.
3792 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3793 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
3794 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3795 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3796 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3797 (aarch64_sqdmull_lane<mode>_internal): Likewise.
3798 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3799
3800 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
3801
3802 * config/aarch64/aarch64-simd.md
3803 (aarch64_be_checked_get_lane<mode>): New define_expand.
3804 * config/aarch64/aarch64-simd-builtins.def
3805 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
3806 New builtin definition.
3807 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
3808 Use new safe be builtin.
3809
3810 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3811
3812 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
3813 New define_insn.
3814 (aarch64_be_st1<mode>): Likewise.
3815 (aarch_ld1<VALL:mode>): Define_expand modified.
3816 (aarch_st1<VALL:mode>): Likewise.
3817 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
3818 (UNSPEC_ST1): Likewise.
3819
3820 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
3821
3822 * config/microblaze/microblaze.md: Add trap insn and attribute
3823
3824 2014-01-23 Dodji Seketeli <dodji@redhat.com>
3825
3826 PR preprocessor/58580
3827 * input.h (location_get_source_line): Take an additional line_size
3828 parameter.
3829 (void diagnostics_file_cache_fini): Declare new function.
3830 * input.c (struct fcache): New type.
3831 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
3832 New static constants.
3833 (diagnostic_file_cache_init, total_lines_num)
3834 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
3835 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
3836 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
3837 (get_next_line, read_next_line, goto_next_line, read_line_num):
3838 New static function definitions.
3839 (diagnostic_file_cache_fini): New function.
3840 (location_get_source_line): Take an additional output line_len
3841 parameter. Re-write using lookup_or_add_file_to_cache_tab and
3842 read_line_num.
3843 * diagnostic.c (diagnostic_finish): Call
3844 diagnostic_file_cache_fini.
3845 (adjust_line): Take an additional input parameter for the length
3846 of the line, rather than calculating it with strlen.
3847 (diagnostic_show_locus): Adjust the use of
3848 location_get_source_line and adjust_line with respect to their new
3849 signature. While displaying a line now, do not stop at the first
3850 null byte. Rather, display the zero byte as a space and keep
3851 going until we reach the size of the line.
3852 * Makefile.in: Add vec.o to OBJS-libcommon
3853
3854 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3855 Ilya Tocar <ilya.tocar@intel.com>
3856
3857 * config/i386/avx512fintrin.h (_mm512_kmov): New.
3858 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
3859 (__builtin_ia32_kmov16): Ditto.
3860 * config/i386/i386.md (UNSPEC_KMOV): New.
3861 (kmovw): Ditto.
3862
3863 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3864
3865 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
3866 (_mm512_storeu_si512): Ditto.
3867
3868 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
3869
3870 PR target/52125
3871 * rtl.h (get_referenced_operands): Declare.
3872 * recog.c (get_referenced_operands): New function.
3873 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
3874 operands have been referenced when recording LO_SUM references.
3875
3876 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
3877
3878 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
3879
3880 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3881
3882 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
3883 Enable for generic and recent AMD targets.
3884
3885 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3886
3887 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
3888 ARG_SIZE note when adjustment was eliminated.
3889
3890 2014-01-22 Jeff Law <law@redhat.com>
3891
3892 PR tree-optimization/59597
3893 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
3894 in file. Accept new argument REGISTERING and use it to modify
3895 dump output appropriately.
3896 (register_jump_thread): Corresponding changes.
3897 (mark_threaded_blocks): Reinstate code to cancel unprofitable
3898 thread paths involving joiner blocks. Add code to dump cancelled
3899 jump threading paths.
3900
3901 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
3902
3903 PR rtl-optimization/59477
3904 * lra-constraints.c (inherit_in_ebb): Process call for living hard
3905 regs. Update reloads_num and potential_reload_hard_regs for all insns.
3906
3907 2014-01-22 Tom Tromey <tromey@redhat.com>
3908
3909 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
3910 PARAMS.
3911 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
3912
3913 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3914
3915 PR rtl-optimization/59896
3916 * lra-constraints.c (process_alt_operands): Check unused note for
3917 matched operands of insn with no output reloads.
3918
3919 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
3920
3921 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
3922 (mips_move_from_gpr_cost): Likewise.
3923
3924 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3925
3926 PR rtl-optimization/59858
3927 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
3928 ira_class_hard_regs_num.
3929 (process_alt_operands): Increase reject for dying matched operand.
3930
3931 2014-01-21 Jakub Jelinek <jakub@redhat.com>
3932
3933 PR target/59003
3934 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
3935 smaller than size, perform several stores or loads and stores
3936 at dst + count - size to store or copy all of size bytes, rather
3937 than just last modesize bytes.
3938
3939 2014-01-20 DJ Delorie <dj@redhat.com>
3940
3941 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
3942 that CLOBBERs are REGs before propogating their values.
3943
3944 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
3945
3946 PR middle-end/59789
3947 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
3948 (cgraph_inline_failed_type): New function.
3949 * cgraph.h (DEFCIFCODE): Add type.
3950 (cgraph_inline_failed_type_t): New enum.
3951 (cgraph_inline_failed_type): New prototype.
3952 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
3953 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
3954 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
3955 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
3956 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
3957 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
3958 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
3959 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
3960 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
3961 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
3962 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
3963 OPTIMIZATION_MISMATCH.
3964 * tree-inline.c (expand_call_inline): Emit errors during
3965 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
3966
3967 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3968
3969 PR target/59685
3970 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
3971 mode attribute in insn output.
3972
3973 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
3974
3975 * output.h (output_constant): Delete.
3976 * varasm.c (output_constant): Make private.
3977
3978 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
3979
3980 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
3981
3982 2014-01-20 Jakub Jelinek <jakub@redhat.com>
3983
3984 PR middle-end/59860
3985 * tree.h (fold_builtin_strcat): New prototype.
3986 * builtins.c (fold_builtin_strcat): No longer static. Add len
3987 argument, if non-NULL, don't call c_strlen. Optimize
3988 directly into __builtin_memcpy instead of __builtin_strcpy.
3989 (fold_builtin_2): Adjust fold_builtin_strcat caller.
3990 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
3991
3992 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3993
3994 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
3995 for SImode_address_operand operands, having only a REG argument.
3996
3997 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
3998
3999 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
4000 loader name using mbig-endian.
4001 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
4002
4003 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4004
4005 * doc/invoke.texi (-march): Clarify documentation for AArch64.
4006 (-mtune): Likewise.
4007 (-mcpu): Likewise.
4008
4009 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
4010
4011 * config/aarch64/aarch64-protos.h
4012 (aarch64_cannot_change_mode_class_ptr): Declare.
4013 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
4014 aarch64_cannot_change_mode_class_ptr): New.
4015 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
4016 backend hook aarch64_cannot_change_mode_class.
4017
4018 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4019
4020 * common/config/aarch64/aarch64-common.c
4021 (aarch64_handle_option): Don't handle any option order logic here.
4022 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
4023 selected_cpu, warn on architecture version mismatch.
4024 (aarch64_override_options): Fix parsing order for option strings.
4025
4026 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4027 Iain Sandoe <iain@codesourcery.com>
4028
4029 PR bootstrap/59496
4030 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
4031 warning. Amend comment to reflect current functionality.
4032
4033 2014-01-20 Richard Biener <rguenther@suse.de>
4034
4035 PR middle-end/59860
4036 * builtins.c (fold_builtin_strcat): Remove case better handled
4037 by tree-ssa-strlen.c.
4038
4039 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
4040
4041 * config/aarch64/aarch64.opt
4042 (mcpu, march, mtune): Make case-insensitive.
4043
4044 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4045
4046 PR target/59880
4047 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4048 if operands[1] is a REG or ZERO_EXTEND of a REG.
4049
4050 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
4051
4052 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
4053
4054 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
4055
4056 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
4057 long non-pic millicode calls.
4058
4059 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4060
4061 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
4062
4063 2014-01-19 Kito Cheng <kito@0xlab.org>
4064
4065 * builtins.c (expand_movstr): Check movstr expand done or fail.
4066
4067 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4068 H.J. Lu <hongjiu.lu@intel.com>
4069
4070 PR target/59379
4071 * config/i386/i386.md (*lea<mode>): Zero-extend return register
4072 to DImode for zero-extended addresses.
4073
4074 2014-01-19 Jakub Jelinek <jakub@redhat.com>
4075
4076 PR rtl-optimization/57763
4077 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
4078 on the new indirect jump_insn and increment LABEL_NUSES (label).
4079
4080 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
4081
4082 PR bootstrap/59580
4083 PR bootstrap/59583
4084 * config.gcc (x86_archs): New variable.
4085 (x86_64_archs): Likewise.
4086 (x86_cpus): Likewise.
4087 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
4088 --with-arch/--with-cpu= options.
4089 Support --with-arch=/--with-cpu={nehalem,westmere,
4090 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
4091
4092 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4093
4094 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
4095 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
4096
4097 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4098
4099 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
4100
4101 2014-01-18 Jakub Jelinek <jakub@redhat.com>
4102
4103 PR target/58944
4104 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
4105 clear cpp_get_options (parse_in)->warn_unused_macros for
4106 ix86_target_macros_internal with cpp_define.
4107
4108 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
4109
4110 * jump.c (delete_related_insns): Keep (use (insn))s.
4111 * reorg.c (redundant_insn): Check for barriers too.
4112
4113 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4114
4115 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
4116
4117 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
4118
4119 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
4120 call to $$dyncall when TARGET_LONG_CALLS is true.
4121
4122 2014-01-17 Jeff Law <law@redhat.com>
4123
4124 * ree.c (combine_set_extension): Temporarily disable test for
4125 changing number of hard registers.
4126
4127 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4128
4129 PR middle-end/58125
4130 * ipa-inline-analysis.c (inline_free_summary):
4131 Do not free summary of aliases.
4132
4133 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4134
4135 PR middle-end/59706
4136 * gimplify.c (gimplify_expr): Use create_tmp_var
4137 instead of create_tmp_var_raw. If cond doesn't have
4138 integral type, don't add the IFN_ANNOTATE builtin at all.
4139
4140 2014-01-17 Martin Jambor <mjambor@suse.cz>
4141
4142 PR ipa/59736
4143 * ipa-cp.c (prev_edge_clone): New variable.
4144 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
4145 Also resize prev_edge_clone vector.
4146 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
4147 (ipcp_edge_removal_hook): New function.
4148 (ipcp_driver): Register ipcp_edge_removal_hook.
4149
4150 2014-01-17 Andrew Pinski <apinski@cavium.com>
4151 Steve Ellcey <sellcey@mips.com>
4152
4153 PR target/59462
4154 * config/mips/mips.c (mips_print_operand): Check operand mode instead
4155 of operator mode.
4156
4157 2014-01-17 Jeff Law <law@redhat.com>
4158
4159 PR middle-end/57904
4160 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
4161 so that pass_ccp runs first.
4162
4163 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4164
4165 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
4166 (ix86_adjust_cost): Use !TARGET_XXX.
4167 (do_reorder_for_imul): Likewise.
4168 (swap_top_of_ready_list): Likewise.
4169 (ix86_sched_reorder): Likewise.
4170
4171 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4172
4173 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4174 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
4175 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
4176 (intel_memset): New. Duplicate slm_memset.
4177 (intel_cost): New. Duplicate slm_cost.
4178 (m_INTEL): New macro.
4179 (processor_target_table): Add "intel".
4180 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
4181 with PROCESSOR_INTEL for "intel".
4182 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
4183 PROCESSOR_SILVERMONT.
4184 (ix86_issue_rate): Likewise.
4185 (ix86_adjust_cost): Likewise.
4186 (ia32_multipass_dfa_lookahead): Likewise.
4187 (swap_top_of_ready_list): Likewise.
4188 (ix86_sched_reorder): Likewise.
4189 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
4190 instead of TARGET_OPT_AGU.
4191 * config/i386/i386.h (TARGET_INTEL): New.
4192 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
4193 (processor_type): Add PROCESSOR_INTEL.
4194 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
4195 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
4196
4197 2014-01-17 Marek Polacek <polacek@redhat.com>
4198
4199 PR c/58346
4200 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
4201 size is zero.
4202
4203 2014-01-17 Richard Biener <rguenther@suse.de>
4204
4205 PR tree-optimization/46590
4206 * opts.c (default_options_table): Add entries for
4207 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
4208 all enabled at -O1 but not for -Og.
4209 * common.opt (fbranch-count-reg): Remove Init(1).
4210 (fmove-loop-invariants): Likewise.
4211 (ftree-pta): Likewise.
4212
4213 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4214
4215 * config/i386/i386.c (ix86_data_alignment): For compatibility with
4216 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
4217 decls to at least the GCC 4.8 used alignments.
4218
4219 PR fortran/59440
4220 * tree-nested.c (convert_nonlocal_reference_stmt,
4221 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
4222 of GIMPLE_BIND stmts, adjust associated decls.
4223
4224 2014-01-17 Richard Biener <rguenther@suse.de>
4225
4226 PR tree-optimization/46590
4227 * vec.h (vec<>::bseach): New member function implementing
4228 binary search according to C89 bsearch.
4229 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
4230 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
4231 bitmap pointer again. Make accesses_in_loop a flat array.
4232 (mem_ref_obstack): New global.
4233 (outermost_indep_loop): Adjust for mem_ref->stored changes.
4234 (mark_ref_stored): Likewise.
4235 (ref_indep_loop_p_2): Likewise.
4236 (set_ref_stored_in_loop): New helper function.
4237 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
4238 (memref_free): Adjust.
4239 (record_mem_ref_loc): Simplify.
4240 (gather_mem_refs_stmt): Adjust.
4241 (sort_locs_in_loop_postorder_cmp): New function.
4242 (analyze_memory_references): Sort accesses_in_loop after
4243 loop postorder number.
4244 (find_ref_loc_in_loop_cmp): New function.
4245 (for_all_locs_in_loop): Find relevant cluster of locs in
4246 accesses_in_loop and iterate without recursion.
4247 (execute_sm): Avoid uninit warning.
4248 (struct ref_always_accessed): Simplify.
4249 (ref_always_accessed::operator ()): Likewise.
4250 (ref_always_accessed_p): Likewise.
4251 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
4252 loop postorder numbers here.
4253 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
4254 numbers.
4255
4256 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4257
4258 PR c++/57945
4259 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
4260 on decls for which assemble_alias has been called.
4261
4262 2014-01-17 Nick Clifton <nickc@redhat.com>
4263
4264 * config/msp430/msp430.opt: (mcpu): New option.
4265 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
4266 (msp430_option_override): Parse target_cpu. If the MCU name
4267 matches a generic string, clear target_mcu.
4268 (msp430_attr): Allow numeric interrupt values up to 63.
4269 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
4270 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
4271 option.
4272 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
4273 Add mcpu matches.
4274 * config/msp430/msp430.md (popm): Use %J rather than %I.
4275 (addsi3): Use msp430_nonimmediate_operand for operand 2.
4276 (addhi_cy_i): Use immediate_operand for operand 2.
4277 * doc/invoke.texi: Document -mcpu option.
4278
4279 2014-01-17 Richard Biener <rguenther@suse.de>
4280
4281 PR rtl-optimization/38518
4282 * df.h (df_analyze_loop): Declare.
4283 * df-core.c: Include cfgloop.h.
4284 (df_analyze_1): Split out main part of df_analyze.
4285 (df_analyze): Adjust.
4286 (loop_inverted_post_order_compute): New function.
4287 (loop_post_order_compute): Likewise.
4288 (df_analyze_loop): New function avoiding whole-function
4289 postorder computes.
4290 * loop-invariant.c (find_defs): Use df_analyze_loop.
4291 (find_invariants): Adjust.
4292 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
4293
4294 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
4295
4296 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
4297 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
4298
4299 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
4300
4301 * ipa-ref.c (ipa_remove_stmt_references): Fix references
4302 traversal when removing references.
4303
4304 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
4305
4306 PR ipa/59775
4307 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
4308
4309 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
4310
4311 PR middle-end/56791
4312 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
4313 pushing a reload for an autoinc when we had previously reloaded an
4314 inner part of the address.
4315
4316 2014-01-16 Jakub Jelinek <jakub@redhat.com>
4317
4318 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
4319 field.
4320 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
4321 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
4322 when not giving up or versioning for alias only because of
4323 loop->safelen.
4324 (vect_analyze_data_ref_dependences): Set to true.
4325 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
4326 is a GIMPLE_PHI.
4327 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
4328 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
4329 to the condition.
4330
4331 PR middle-end/58344
4332 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
4333
4334 PR target/59839
4335 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
4336 operand 0 predicate for gathers, use a new pseudo as subtarget.
4337
4338 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4339
4340 PR middle-end/59609
4341 * lra-constraints.c (process_alt_operands): Add printing debug info.
4342 Check absence of input/output reloads for matched operands too.
4343
4344 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4345
4346 PR rtl-optimization/59835
4347 * ira.c (ira_init_register_move_cost): Increase cost for
4348 impossible modes.
4349
4350 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
4351
4352 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
4353
4354 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
4355
4356 PR target/59780
4357 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
4358 non-register objects. Use gen_(high/low)part more consistently.
4359 Fix assertions.
4360
4361 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
4362
4363 PR target/59844
4364 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
4365 endian support, remove tests for WORDS_BIG_ENDIAN.
4366 (p8_mfvsrd_3_<mode>): Likewise.
4367 (reload_gpr_from_vsx<mode>): Likewise.
4368 (reload_gpr_from_vsxsf): Likewise.
4369 (p8_mfvsrd_4_disf): Likewise.
4370
4371 2014-01-16 Richard Biener <rguenther@suse.de>
4372
4373 PR rtl-optimization/46590
4374 * lcm.c (compute_antinout_edge): Use postorder iteration.
4375 (compute_laterin): Use inverted postorder iteration.
4376
4377 2014-01-16 Nick Clifton <nickc@redhat.com>
4378
4379 PR middle-end/28865
4380 * varasm.c (output_constant): Return the number of bytes actually
4381 emitted.
4382 (output_constructor_array_range): Update the field size with the
4383 number of bytes emitted by output_constant.
4384 (output_constructor_regular_field): Likewise. Also do not
4385 complain if the total number of bytes emitted is now greater
4386 than the expected fieldpos.
4387 * output.h (output_constant): Update prototype and descriptive comment.
4388
4389 2014-01-16 Marek Polacek <polacek@redhat.com>
4390
4391 PR middle-end/59827
4392 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
4393 it is error_mark_node.
4394
4395 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
4396
4397 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
4398 VALID_AVX256_REG_OR_OI_MODE.
4399
4400 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
4401
4402 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
4403 current procedure should be profiled.
4404
4405 2014-01-15 Andrew Pinski <apinski@cavium.com>
4406
4407 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
4408 of moving from/to the STACK_REG register class.
4409
4410 2014-01-15 Richard Henderson <rth@redhat.com>
4411
4412 PR debug/54694
4413 * reginfo.c (global_regs_decl): Globalize.
4414 * rtl.h (global_regs_decl): Declare.
4415 * ira.c (do_reload): Diagnose frame_pointer_needed and it
4416 reserved via global_regs.
4417
4418 2014-01-15 Teresa Johnson <tejohnson@google.com>
4419
4420 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
4421
4422 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4423
4424 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
4425 and vmulosh rather than call gen_vec_widen_smult_*.
4426 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
4427 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
4428 (vec_widen_smult_even_v16qi): Likewise.
4429 (vec_widen_umult_even_v8hi): Likewise.
4430 (vec_widen_smult_even_v8hi): Likewise.
4431 (vec_widen_umult_odd_v16qi): Likewise.
4432 (vec_widen_smult_odd_v16qi): Likewise.
4433 (vec_widen_umult_odd_v8hi): Likewise.
4434 (vec_widen_smult_odd_v8hi): Likewise.
4435 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
4436 vmuloub rather than call gen_vec_widen_umult_*.
4437 (vec_widen_umult_lo_v16qi): Likewise.
4438 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
4439 vmulosb rather than call gen_vec_widen_smult_*.
4440 (vec_widen_smult_lo_v16qi): Likewise.
4441 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
4442 rather than call gen_vec_widen_umult_*.
4443 (vec_widen_umult_lo_v8hi): Likewise.
4444 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
4445 rather than call gen_vec_widen_smult_*.
4446 (vec_widen_smult_lo_v8hi): Likewise.
4447
4448 2014-01-15 Jeff Law <law@redhat.com>
4449
4450 PR tree-optimization/59747
4451 * ree.c (find_and_remove_re): Properly handle case where a second
4452 eliminated extension requires widening a copy created for elimination
4453 of a prior extension.
4454 (combine_set_extension): Ensure that the number of hard regs needed
4455 for a destination register does not change when we widen it.
4456
4457 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
4458
4459 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
4460 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
4461 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
4462 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
4463 (avr-*-rtems*): Likewise.
4464 (bfin*-rtems*): Likewise.
4465 (moxie-*-rtems*): Likewise.
4466 (h8300-*-rtems*): Likewise.
4467 (i[34567]86-*-rtems*): Likewise.
4468 (lm32-*-rtems*): Likewise.
4469 (m32r-*-rtems*): Likewise.
4470 (m68k-*-rtems*): Likewise.
4471 (microblaze*-*-rtems*): Likewise.
4472 (mips*-*-rtems*): Likewise.
4473 (powerpc-*-rtems*): Likewise.
4474 (sh-*-rtems*): Likewise.
4475 (sparc-*-rtems*): Likewise.
4476 (sparc64-*-rtems*): Likewise.
4477 (v850-*-rtems*): Likewise.
4478 (m32c-*-rtems*): Likewise.
4479
4480 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
4481
4482 PR rtl-optimization/59511
4483 * ira.c (ira_init_register_move_cost): Use memory costs for some
4484 cases of register move cost calculations.
4485 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
4486 instead of BB frequency.
4487 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
4488 * lra-assigns.c (find_hard_regno_for): Ditto.
4489
4490 2014-01-15 Richard Biener <rguenther@suse.de>
4491
4492 PR tree-optimization/59822
4493 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
4494 (vectorizable_load): Use it to hoist defs of uses of invariant
4495 loads out of the loop.
4496
4497 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
4498 Kugan Vivekanandarajah <kuganv@linaro.org>
4499
4500 PR target/59695
4501 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
4502 truncation.
4503
4504 2014-01-15 Richard Biener <rguenther@suse.de>
4505
4506 PR rtl-optimization/59802
4507 * lcm.c (compute_available): Use inverted postorder to seed
4508 the initial worklist.
4509
4510 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4511
4512 PR target/59803
4513 * config/s390/s390.c (s390_preferred_reload_class): Don't return
4514 ADDR_REGS for invalid symrefs in non-PIC code.
4515
4516 2014-01-15 Jakub Jelinek <jakub@redhat.com>
4517
4518 PR other/58712
4519 * builtins.c (determine_block_size): Initialize *probable_max_size
4520 even if len_rtx is CONST_INT.
4521
4522 2014-01-14 Andrew Pinski <apinski@cavium.com>
4523
4524 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
4525 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
4526 (cortexa53_tunings): Likewise.
4527 (aarch64_sched_issue_rate): New function.
4528 (TARGET_SCHED_ISSUE_RATE): Define.
4529
4530 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4531
4532 * ira-costs.c (find_costs_and_classes): Add missed
4533 ira_init_register_move_cost_if_necessary.
4534
4535 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4536
4537 PR target/59787
4538 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
4539
4540 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
4541
4542 PR target/59794
4543 * config/i386/i386.c (type_natural_mode): Add a bool parameter
4544 to indicate if type is used for function return value. Warn ABI
4545 change if the vector mode isn't available for function return value.
4546 (ix86_function_arg_advance): Pass false to type_natural_mode.
4547 (ix86_function_arg): Likewise.
4548 (ix86_gimplify_va_arg): Likewise.
4549 (function_arg_32): Don't warn ABI change.
4550 (ix86_function_value): Pass true to type_natural_mode.
4551 (ix86_return_in_memory): Likewise.
4552 (ix86_struct_value_rtx): Removed.
4553 (TARGET_STRUCT_VALUE_RTX): Likewise.
4554
4555 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4556
4557 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
4558 converting a conditional jump into a conditional return.
4559
4560 2014-01-14 Richard Biener <rguenther@suse.de>
4561
4562 PR tree-optimization/58921
4563 PR tree-optimization/59006
4564 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
4565 hoisting invariant stmts.
4566 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
4567 invariant loads on the preheader edge if possible.
4568
4569 2014-01-14 Joey Ye <joey.ye@arm.com>
4570
4571 * doc/plugin.texi (Building GCC plugins): Update to C++.
4572
4573 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
4574
4575 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
4576 (_mm_rcp28_round_ss): Ditto.
4577 (_mm_rsqrt28_round_sd): Ditto.
4578 (_mm_rsqrt28_round_ss): Ditto.
4579 (_mm_rcp28_sd): Ditto.
4580 (_mm_rcp28_ss): Ditto.
4581 (_mm_rsqrt28_sd): Ditto.
4582 (_mm_rsqrt28_ss): Ditto.
4583 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
4584 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
4585 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
4586 (IX86_BUILTIN_RCP28SD): Ditto.
4587 (IX86_BUILTIN_RCP28SS): Ditto.
4588 (IX86_BUILTIN_RSQRT28SD): Ditto.
4589 (IX86_BUILTIN_RSQRT28SS): Ditto.
4590 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
4591 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
4592 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
4593 (ix86_expand_special_args_builtin): Expand new FTYPE.
4594 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
4595 (srcp14<mode>): Make insn unary.
4596 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
4597 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
4598 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
4599 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
4600 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
4601 Fix rounding: make it SAE only.
4602 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
4603 Ditto.
4604 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
4605 Ditto.
4606 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
4607 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
4608 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
4609 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
4610 (round_saeonly_mask_scalar_operand4): Ditto.
4611 (round_saeonly_mask_scalar_op3): Ditto.
4612 (round_saeonly_mask_scalar_op4): Ditto.
4613
4614 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4615
4616 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4617 Implement -maltivec=be for vec_insert and vec_extract.
4618
4619 2014-01-10 DJ Delorie <dj@redhat.com>
4620
4621 * config/msp430/msp430.md (call_internal): Don't allow memory
4622 references with SP as the base register.
4623 (call_value_internal): Likewise.
4624 * config/msp430/constraints.md (Yc): New. For memory references
4625 that don't use SP as a base register.
4626
4627 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
4628 "an integer without a # prefix"
4629 * config/msp430/msp430.md (epilogue_helper): Use it.
4630
4631 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4632
4633 PR target/59617
4634 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
4635 AVX512F gather builtins.
4636 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
4637 on gather decls with INTEGER_TYPE masktype.
4638 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
4639 directly into the builtin rather than hoisting it before loop.
4640
4641 PR tree-optimization/59387
4642 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
4643 (scev_const_prop): If folded_casts and type has undefined overflow,
4644 use force_gimple_operand instead of force_gimple_operand_gsi and
4645 for each added stmt if it is assign with
4646 arith_code_with_undefined_signed_overflow, call
4647 rewrite_to_defined_overflow.
4648 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
4649 gimple-fold.h instead.
4650 (arith_code_with_undefined_signed_overflow,
4651 rewrite_to_defined_overflow): Moved to ...
4652 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
4653 rewrite_to_defined_overflow): ... here. No longer static.
4654 Include gimplify-me.h.
4655 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
4656 rewrite_to_defined_overflow): New prototypes.
4657
4658 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4659
4660 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
4661
4662 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
4663
4664 * builtins.c (get_object_alignment_2): Minor tweak.
4665 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
4666
4667 2014-01-13 Christian Bruel <christian.bruel@st.com>
4668
4669 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
4670 optimized non constant lengths.
4671
4672 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4673
4674 PR libgomp/59194
4675 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
4676 load as __atomic_load_N if possible.
4677
4678 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4679
4680 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
4681 target parameter.
4682 (rs6000_expand_builtin): Adjust call.
4683
4684 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4685
4686 PR target/58115
4687 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
4688 * config/rs6000/rs6000.c: Include target-globals.h.
4689 (rs6000_set_current_function): Instead of doing target_reinit
4690 unconditionally, use save_target_globals_default_opts and
4691 restore_target_globals.
4692
4693 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
4694 FPSCR.
4695 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
4696 (rs6000_expand_builtin): Handle mffs and mtfsf.
4697 (rs6000_init_builtins): Define mffs and mtfsf.
4698 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
4699 (rs6000_mffs): New pattern.
4700 (rs6000_mtfsf): New pattern.
4701
4702 2014-01-11 Bin Cheng <bin.cheng@arm.com>
4703
4704 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
4705 Start narrowing with START. Apply candidate-use pair
4706 and check overall cost in narrowing.
4707 (iv_ca_prune): Pass new argument.
4708
4709 2014-01-10 Jeff Law <law@redhat.com>
4710
4711 PR middle-end/59743
4712 * ree.c (combine_reaching_defs): Ensure the defining statement
4713 occurs before the extension when optimizing extensions with
4714 different source and destination hard registers.
4715
4716 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4717
4718 PR ipa/58585
4719 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
4720 vtables into the type inheritance graph.
4721
4722 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4723
4724 PR rtl-optimization/59754
4725 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
4726 modes in the REGNO != REGNO case.
4727
4728 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4729
4730 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
4731
4732 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4733
4734 PR tree-optimization/59745
4735 * tree-predcom.c (tree_predictive_commoning_loop): Call
4736 free_affine_expand_cache if giving up because components is NULL.
4737
4738 * target-globals.c (save_target_globals): Allocate < 4KB structs using
4739 GC in payload of target_globals struct instead of allocating them on
4740 the heap and the larger structs separately using GC.
4741 * target-globals.h (struct target_globals): Make regs, hard_regs,
4742 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
4743 of GTY((skip)) and change type to void *.
4744 (reset_target_globals): Cast loads from those fields to corresponding
4745 types.
4746
4747 2014-01-10 Steve Ellcey <sellcey@mips.com>
4748
4749 PR plugins/59335
4750 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
4751 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
4752 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
4753
4754 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
4755
4756 PR target/59744
4757 * aarch64-modes.def (CC_Zmode): New flags mode.
4758 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
4759 represents an equality.
4760 (aarch64_get_condition_code): Handle CC_Zmode.
4761 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
4762
4763 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4764
4765 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
4766 extraction in good case.
4767
4768 2014-01-10 Richard Biener <rguenther@suse.de>
4769
4770 PR tree-optimization/59374
4771 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
4772 checking after SLP discovery. Mark stmts not participating
4773 in any SLP instance properly.
4774
4775 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4776
4777 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
4778 when handling a SET rtx.
4779
4780 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4781
4782 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
4783 (cortex-a57): Likewise.
4784 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
4785
4786 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4787
4788 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
4789 non-iwmmxt builtins.
4790
4791 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4792
4793 PR ipa/58252
4794 PR ipa/59226
4795 * ipa-devirt.c record_target_from_binfo): Take as argument
4796 stack of binfos and lookup matching one for virtual inheritance.
4797 (possible_polymorphic_call_targets_1): Update.
4798
4799 2014-01-10 Huacai Chen <chenhc@lemote.com>
4800
4801 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
4802 kernel strings for Loongson-2E/2F/3A.
4803
4804 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4805
4806 PR middle-end/59670
4807 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
4808 is_gimple_call before calling gimple_call_internal_p.
4809
4810 2014-01-09 Steve Ellcey <sellcey@mips.com>
4811
4812 * Makefile.in (TREE_FLOW_H): Remove.
4813 (TREE_SSA_H): Add file names from tree-flow.h.
4814 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
4815 * tree.h: Remove tree-flow.h reference.
4816 * hash-table.h: Remove tree-flow.h reference.
4817 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
4818 reference with tree-ssa-loop.h.
4819
4820 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4821
4822 * doc/invoke.texi: Add -maltivec={be,le} options, and document
4823 default element-order behavior for -maltivec.
4824 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
4825 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
4826 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
4827 when targeting big endian, at least for now.
4828 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
4829
4830 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4831
4832 PR middle-end/47735
4833 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
4834 var satisfies use_register_for_decl, just take into account type
4835 alignment, rather than decl alignment.
4836
4837 PR tree-optimization/59622
4838 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
4839 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
4840 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
4841 Don't devirtualize for inplace at all. For targets.length () == 1,
4842 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
4843
4844 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4845
4846 * config/i386/i386.md (cpu): Remove the unused btver1.
4847
4848 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4849
4850 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
4851
4852 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4853
4854 PR target/58115
4855 * tree-core.h (struct target_globals): New forward declaration.
4856 (struct tree_target_option): Add globals field.
4857 * tree.h (TREE_TARGET_GLOBALS): Define.
4858 (prepare_target_option_nodes_for_pch): New prototype.
4859 * target-globals.h (struct target_globals): Define even if
4860 !SWITCHABLE_TARGET.
4861 * tree.c (prepare_target_option_node_for_pch,
4862 prepare_target_option_nodes_for_pch): New functions.
4863 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
4864 * config/i386/i386.c: Include target-globals.h.
4865 (ix86_set_current_function): Instead of doing target_reinit
4866 unconditionally, use save_target_globals_default_opts and
4867 restore_target_globals.
4868
4869 2014-01-09 Richard Biener <rguenther@suse.de>
4870
4871 PR tree-optimization/59715
4872 * tree-cfg.h (split_critical_edges): Declare.
4873 * tree-cfg.c (split_critical_edges): Export.
4874 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
4875
4876 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
4877
4878 * cfgexpand.c (expand_stack_vars): Optionally disable
4879 asan stack protection.
4880 (expand_used_vars): Likewise.
4881 (partition_stack_vars): Likewise.
4882 * asan.c (asan_emit_stack_protection): Optionally disable
4883 after return stack usage.
4884 (instrument_derefs): Optionally disable memory access instrumentation.
4885 (instrument_builtin_call): Likewise.
4886 (instrument_strlen_call): Likewise.
4887 (asan_protect_global): Optionally disable global variables protection.
4888 * doc/invoke.texi: Added doc for new options.
4889 * params.def: Added new options.
4890 * params.h: Likewise.
4891
4892 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4893
4894 PR rtl-optimization/59724
4895 * ifcvt.c (cond_exec_process_if_block): Don't call
4896 flow_find_head_matching_sequence with 0 longest_match.
4897 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
4898 non-active insns if !stop_after.
4899 (try_head_merge_bb): Revert 2014-01-07 changes.
4900
4901 2014-01-08 Jeff Law <law@redhat.com>
4902
4903 * ree.c (get_sub_rtx): New function, extracted from...
4904 (merge_def_and_ext): Here.
4905 (combine_reaching_defs): Use get_sub_rtx.
4906
4907 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
4908
4909 * cgraph.h (varpool_variable_node): Do not choke on null node.
4910
4911 2014-01-08 Catherine Moore <clm@codesourcery.com>
4912
4913 * config/mips/mips.md (simple_return): Attempt to use JRC
4914 for microMIPS.
4915 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
4916
4917 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4918
4919 PR rtl-optimization/59137
4920 * reorg.c (steal_delay_list_from_target): Call update_block for
4921 elided insns.
4922 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
4923
4924 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4925
4926 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
4927 two duplicate entries.
4928
4929 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4930
4931 Revert:
4932 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
4933
4934 * config/mips/mips.c (mips_truncated_op_cost): New function.
4935 (mips_rtx_costs): Adjust test for BADDU.
4936 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
4937
4938 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
4939
4940 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
4941 (*baddu_si): ...this new pattern.
4942
4943 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4944
4945 PR ipa/59722
4946 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
4947
4948 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
4949
4950 PR middle-end/57748
4951 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
4952 inner_reference_p.
4953 (expand_expr, expand_normal): Adjust.
4954 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
4955 inner_reference_p. Use inner_reference_p to expand inner references.
4956 (store_expr): Adjust.
4957 * cfgexpand.c (expand_call_stmt): Adjust.
4958
4959 2014-01-08 Rong Xu <xur@google.com>
4960
4961 * gcov-io.c (gcov_var): Move from gcov-io.h.
4962 (gcov_position): Ditto.
4963 (gcov_is_error): Ditto.
4964 (gcov_rewrite): Ditto.
4965 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
4966 only part to libgcc/libgcov.h.
4967
4968 2014-01-08 Marek Polacek <polacek@redhat.com>
4969
4970 PR middle-end/59669
4971 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
4972
4973 2014-01-08 Marek Polacek <polacek@redhat.com>
4974
4975 PR sanitizer/59667
4976 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
4977
4978 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4979
4980 PR rtl-optimization/59649
4981 * stor-layout.c (get_mode_bounds): For BImode return
4982 0 and STORE_FLAG_VALUE.
4983
4984 2014-01-08 Richard Biener <rguenther@suse.de>
4985
4986 PR middle-end/59630
4987 * gimple.h (is_gimple_builtin_call): Remove.
4988 (gimple_builtin_call_types_compatible_p): New.
4989 (gimple_call_builtin_p): New overload.
4990 * gimple.c (is_gimple_builtin_call): Remove.
4991 (validate_call): Rename to ...
4992 (gimple_builtin_call_types_compatible_p): ... this and export. Also
4993 check return types.
4994 (validate_type): New static function.
4995 (gimple_call_builtin_p): New overload and adjust.
4996 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
4997 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
4998 (gimple_fold_stmt_to_constant_1): Likewise.
4999 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
5000
5001 2014-01-08 Richard Biener <rguenther@suse.de>
5002
5003 PR middle-end/59471
5004 * gimplify.c (gimplify_expr): Gimplify register-register type
5005 VIEW_CONVERT_EXPRs to separate stmts.
5006
5007 2014-01-07 Jeff Law <law@redhat.com>
5008
5009 PR middle-end/53623
5010 * ree.c (combine_set_extension): Handle case where source
5011 and destination registers in an extension insn are different.
5012 (combine_reaching_defs): Allow source and destination registers
5013 in extension to be different under limited circumstances.
5014 (add_removable_extension): Remove restriction that the
5015 source and destination registers in the extension are the same.
5016 (find_and_remove_re): Emit a copy from the extension's
5017 destination to its source after the defining insn if
5018 the source and destination registers are different.
5019
5020 PR middle-end/59285
5021 * ifcvt.c (merge_if_block): If we are merging a block with more than
5022 one successor with a block with no successors, remove any BARRIER
5023 after the second block.
5024
5025 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
5026
5027 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
5028
5029 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
5030
5031 PR target/59652
5032 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
5033 for 14-bit register offsets when INT14_OK_STRICT is false.
5034
5035 2014-01-07 Roland Stigge <stigge@antcom.de>
5036 Michael Meissner <meissner@linux.vnet.ibm.com>
5037
5038 PR 57386/target
5039 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
5040 Only check TFmode for SPE constants. Don't check TImode or TDmode.
5041
5042 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
5043
5044 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
5045 -mcpu.
5046
5047 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
5048
5049 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
5050 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
5051 rtx is const0_rtx or not.
5052
5053 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
5054
5055 PR target/58115
5056 * target-globals.c (save_target_globals): Remove this_fn_optab
5057 handling.
5058 * toplev.c: Include optabs.h.
5059 (target_reinit): Temporarily restore the global options if another
5060 set of options are in force.
5061
5062 2014-01-07 Jakub Jelinek <jakub@redhat.com>
5063
5064 PR rtl-optimization/58668
5065 * cfgcleanup.c (flow_find_cross_jump): Don't count
5066 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
5067 to determine what is counted.
5068 (flow_find_head_matching_sequence): Use active_insn_p to determine
5069 what is counted.
5070 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
5071 counting change.
5072 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
5073 determine what is counted.
5074
5075 PR tree-optimization/59643
5076 * tree-predcom.c (split_data_refs_to_components): If one dr is
5077 read and one write, determine_offset fails and the write isn't
5078 in the bad component, just put the read into the bad component.
5079
5080 2014-01-07 Mike Stump <mikestump@comcast.net>
5081 Jakub Jelinek <jakub@redhat.com>
5082
5083 PR pch/59436
5084 * tree-core.h (struct tree_optimization_option): Change optabs
5085 type from unsigned char * to void *.
5086 * optabs.c (init_tree_optimization_optabs): Adjust
5087 TREE_OPTIMIZATION_OPTABS initialization.
5088
5089 2014-01-06 Jakub Jelinek <jakub@redhat.com>
5090
5091 PR target/59644
5092 * config/i386/i386.h (struct machine_function): Add
5093 no_drap_save_restore field.
5094 * config/i386/i386.c (ix86_save_reg): Use
5095 !cfun->machine->no_drap_save_restore instead of
5096 crtl->stack_realign_needed.
5097 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
5098 this function clears frame_pointer_needed. Set
5099 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
5100 and DRAP reg is needed.
5101
5102 2014-01-06 Marek Polacek <polacek@redhat.com>
5103
5104 PR c/57773
5105 * doc/implement-c.texi: Mention that other integer types are
5106 permitted as bit-field types in strictly conforming mode.
5107
5108 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
5109
5110 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
5111 is newly allocated.
5112
5113 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
5114
5115 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
5116
5117 2014-01-06 Martin Jambor <mjambor@suse.cz>
5118
5119 PR ipa/59008
5120 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
5121 to int.
5122 * ipa-prop.c (ipa_print_node_params): Fix indentation.
5123
5124 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
5125
5126 PR debug/59350
5127 PR debug/59510
5128 * var-tracking.c (add_stores): Preserve the value of the source even if
5129 we don't record the store.
5130
5131 2014-01-06 Terry Guo <terry.guo@arm.com>
5132
5133 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
5134
5135 2014-01-05 Iain Sandoe <iain@codesourcery.com>
5136
5137 PR bootstrap/59541
5138 * config/darwin.c (darwin_function_section): Adjust return values to
5139 correspond to optimisation changes made in r206070.
5140
5141 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
5142
5143 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
5144 from prefetch_block tune setting.
5145 (nocona_cost): Correct size of prefetch block to 64.
5146
5147 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
5148
5149 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
5150 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
5151 used to save the static chain register in the computation of the offset
5152 from which the FP registers need to be restored.
5153
5154 2014-01-04 Jakub Jelinek <jakub@redhat.com>
5155
5156 PR tree-optimization/59519
5157 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
5158 ICE if get_current_def (current_new_name) is already non-NULL, as long
5159 as it is a phi result of some other phi in *new_exit_bb that has
5160 the same argument.
5161
5162 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
5163 or vmovdqu* for misaligned_operand.
5164 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
5165 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
5166 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
5167 aligned_mem for AVX512F masked aligned load and store builtins and for
5168 non-temporal moves.
5169
5170 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
5171
5172 PR tree-optimization/59651
5173 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
5174 Address range for negative step should be added by TYPE_SIZE_UNIT.
5175
5176 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
5177
5178 * config/m68k/m68k.c (handle_move_double): Handle pushes with
5179 overlapping registers also for registers other than the stack pointer.
5180
5181 2014-01-03 Marek Polacek <polacek@redhat.com>
5182
5183 PR other/59661
5184 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
5185 __builtin_FILE.
5186
5187 2014-01-03 Jakub Jelinek <jakub@redhat.com>
5188
5189 PR target/59625
5190 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
5191 asm goto as jump.
5192
5193 * config/i386/i386.md (MODE_SIZE): New mode attribute.
5194 (push splitter): Use <P:MODE_SIZE> instead of
5195 GET_MODE_SIZE (<P:MODE>mode).
5196 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
5197 (mov -1, reg peephole2): Likewise.
5198 * config/i386/sse.md (*mov<mode>_internal,
5199 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
5200 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
5201 *<code><mode>3, *andnot<mode>3<mask_name>,
5202 <mask_codefor><code><mode>3<mask_name>): Likewise.
5203 * config/i386/subst.md (mask_mode512bit_condition,
5204 sd_mask_mode512bit_condition): Likewise.
5205
5206 2014-01-02 Xinliang David Li <davidxl@google.com>
5207
5208 PR tree-optimization/59303
5209 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
5210 (dump_predicates): Better output format.
5211 (pred_equal_p): New function.
5212 (is_neq_relop_p): Ditto.
5213 (is_neq_zero_form_p): Ditto.
5214 (pred_expr_equal_p): Ditto.
5215 (pred_neg_p): Ditto.
5216 (simplify_pred): Ditto.
5217 (simplify_preds_2): Ditto.
5218 (simplify_preds_3): Ditto.
5219 (simplify_preds_4): Ditto.
5220 (simplify_preds): Ditto.
5221 (push_pred): Ditto.
5222 (push_to_worklist): Ditto.
5223 (get_pred_info_from_cmp): Ditto.
5224 (is_degenerated_phi): Ditto.
5225 (normalize_one_pred_1): Ditto.
5226 (normalize_one_pred): Ditto.
5227 (normalize_one_pred_chain): Ditto.
5228 (normalize_preds): Ditto.
5229 (normalize_cond_1): Remove function.
5230 (normalize_cond): Ditto.
5231 (is_gcond_subset_of): Ditto.
5232 (is_subset_of_any): Ditto.
5233 (is_or_set_subset_of): Ditto.
5234 (is_and_set_subset_of): Ditto.
5235 (is_norm_cond_subset_of): Ditto.
5236 (pred_chain_length_cmp): Ditto.
5237 (convert_control_dep_chain_into_preds): Type change.
5238 (find_predicates): Ditto.
5239 (find_def_preds): Ditto.
5240 (destroy_predicates_vecs): Ditto.
5241 (find_matching_predicates_in_rest_chains): Ditto.
5242 (use_pred_not_overlap_with_undef_path_pred): Ditto.
5243 (is_pred_expr_subset): Ditto.
5244 (is_pred_chain_subset_of): Ditto.
5245 (is_included_in): Ditto.
5246 (is_superset_of): Ditto.
5247
5248 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5249
5250 Update copyright years.
5251
5252 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5253
5254 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
5255 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
5256 config/arc/arc.md, config/arc/arc.opt,
5257 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
5258 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
5259 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
5260 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
5261 config/linux-protos.h, config/linux.c, config/winnt-c.c,
5262 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
5263 vtable-verify.c, vtable-verify.h: Use the standard form for the
5264 copyright notice.
5265
5266 2014-01-02 Tobias Burnus <burnus@net-b.de>
5267
5268 * gcc.c (process_command): Update copyright notice dates.
5269 * gcov-dump.c: Ditto.
5270 * gcov.c: Ditto.
5271 * doc/cpp.texi: Bump @copying's copyright year.
5272 * doc/cppinternals.texi: Ditto.
5273 * doc/gcc.texi: Ditto.
5274 * doc/gccint.texi: Ditto.
5275 * doc/gcov.texi: Ditto.
5276 * doc/install.texi: Ditto.
5277 * doc/invoke.texi: Ditto.
5278
5279 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5280
5281 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
5282
5283 2014-01-01 Jakub Jelinek <jakub@redhat.com>
5284
5285 * config/i386/sse.md (*mov<mode>_internal): Guard
5286 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
5287
5288 PR rtl-optimization/59647
5289 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
5290 new_rtx into UNSIGNED_FLOAT rtxes.
5291 \f
5292 Copyright (C) 2014 Free Software Foundation, Inc.
5293
5294 Copying and distribution of this file, with or without modification,
5295 are permitted in any medium without royalty provided the copyright
5296 notice and this notice are preserved.